Good Morning, AI Enthusiasts!

One company is asking for your passport to use a chatbot, another is turning sneakers into GPUs.

TOKEN

Claude Asks for Your ID

👀 What’s Happening: Dario Amodei warned in early 2026 that mistiming data center investment by even a few years could be “ruinous,” taking a swipe at what sounded like YOLO-style buildouts. Months later, Anthropic began gating Claude with ID verification and tighter access, requiring government-issued photo IDs and live camera checks for some use cases.

🌍 How This Hits Reality: The contrast is hard to ignore. OpenAI is tied to projects targeting up to $500 billion in AI infrastructure and multi-gigawatt capacity, with early deployments expected by 2026. Anthropic, backed by Amazon and Google, has announced large-scale TPU capacity, but meaningful new supply lands closer to 2027. In the meantime, demand for agent-style workloads is already multiplying token usage severalfold.

🛎️ Key Takeaway: The irony is structural. Calling excess early sounds disciplined, until scarcity shows up in your own product. If supply lags demand, the rhetoric flips fast from caution to constraint, and users feel it first.

TOGETHER WITH MYCLAW

Personal AI That Starts Personal

A personal AI shouldn’t feel generic. MyClaw lets you install skills that adapt to your preferences from the start, so your assistant feels relevant on day one. Behind the scenes, OpenClaw provides the agent power. What you see is a simpler way to put it to work.

If you’re serious about OpenClaw, don’t run it halfway. Run it the way it was meant to be run — fast, secure, and production-ready by default.

IPHONE

Apple Is Kicking AI Builders Out

👀 What’s Happening: Apple has started blocking updates from vibe Coding apps like Replit and Anything, claiming that AI-generated code bypasses App Store review and creates security risks. In reality, these tools mostly generate cloud-based web apps, not malware. Apple is forcing them to weaken features, delaying updates, and leaving iPhone users stuck on outdated versions.

🌍 How This Hits Reality: Apple is trying to protect the App Store’s control over distribution, payments, and its 15% to 30% commission. But Vibe Coding changes who gets to build software. People no longer need Xcode, a Mac, or years of engineering knowledge. They can create products directly from a browser or phone. By fighting this shift, Apple is trying to maintain the same closed-platform logic that once let Windows dominate the PC era, which is impossible.

🛎️ Key Takeaway: Apple is turning its own ecosystem into a trap that only traps itself. The harder it blocks AI-native tools, the faster developers and users will move toward the open web, Android, and OpenClaw like platforms that let them build without asking permission first.

INFRA

Allbirds Just Became an AI Infra

👀 What’s Happening: Once known for wool sneakers and Silicon Valley branding, Allbirds just sold off its shoe business, raised $50 million in convertible debt, and said it will pivot into GPU rental infrastructure under the new name “NewBird AI.” The market loved it. Shares briefly surged more than 800% in a single day. What was recently a struggling shoe company with a valuation only near $39 million suddenly became an AI infrastructure play overnight.

🌍 How This Hits Reality: This is what AI mania looks like. A consumer brand with no real history in chips, data centers, or cloud infrastructure can attach itself to GPU demand and instantly create more paper value than years of selling physical products. Investors are now treating “AI compute” as a stronger story than revenue, margins, supply chains, or product-market fit. The story is worth more than the business that built it.

🛎️ Key Takeaway: Expect more dying public companies to attempt the same move. AI infrastructure may become the new reverse merger trade.

AGENTS

Starbucks Releases a Skill

👀 What’s Happening: Starbucks launched a beta app inside ChatGPT that lets users describe a mood, craving, or dietary preference and get drink suggestions instantly without using menus or search bars. Users can customize the order and choose a nearby store, then finish checkout inside Starbucks’ own app. Right now it still stops short of full in-chat payment, but the connection layer is now in place.

🌍 How This Hits Reality: This matters because Starbucks is no longer acting like just a coffee chain. It is acting like a service endpoint for AI agents. Today ChatGPT suggests a drink and hands you off. Tomorrow, agents like OpenClaw could remember your usual order, your lactose preference, your location, your schedule, and directly place the order before your 9 a.m. meeting. The important shift is that brands are starting to expose themselves to AI workflows instead of forcing users through menus and search bars.

🛎️ Key Takeaway: This is an early but important move toward agent commerce. The winners may not be the brands with the best apps. They may be the brands that are easiest for AI systems to plug into and transact with.

DAILY TL;DR

• OpenAI updated its Agents SDK with sandboxing and long-horizon task support to help enterprises build safer and more capable agents.
• Adobe is bringing its Firefly creative agent into Claude, allowing users to move from idea generation to image editing directly inside the assistant.
• Anthropic’s Project Glasswing may have uncovered around 40 vulnerabilities, but only one FreeBSD flaw can currently be directly tied to the program.
• Google launched a Gemini app for Mac that lets users access the AI assistant anywhere on the desktop and share on-screen content for context.
• LinkedIn says hiring declines are still driven more by the economy than AI, but AI is expected to reshape the skills required for most jobs.
• AI learning app Gizmo has grown to 13 million users by turning notes into gamified study content and has now raised $22 million.
• Accel raised $5 billion to double down on late-stage AI investments across software, robotics, defense, and data center infrastructure.
• Jane Street signed a $6 billion cloud deal with CoreWeave and added a $1 billion investment, pushing its valuation, contracts, and spending even higher.
• Cadence and Nvidia are combining physics simulation engines with robot AI models while also expanding AI’s role in chip design.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.

Good Morning, AI Enthusiasts!

The models are learning to hack companies, fix quantum computers, drive taxis, and win benchmarks nobody trusts.

TEST

Mythos May Be the First Real AI Nuclear Bomb

👀 What's happening: Claude Mythos Preview became the first AI model to fully complete a realistic enterprise network attack simulation built by the UK AI Security Institute. In the test, it autonomously completed a 32-step attack chain that normally takes human experts around 20 hours. It scanned networks, found vulnerabilities, escalated privileges, moved laterally between machines, and eventually took over the entire environment without human intervention.

🌍 How this hits reality: This feels like the Oppenheimer moment for cybersecurity. For years, AI hacking meant writing code snippets or helping humans search for exploits. Claude Mythos Preview crossed into something else. It can now operate like a full offensive team with planning, adaptation, persistence, and retries. Most companies still run on old software, weak internal controls, exposed admin tools, and employees who click on everything. A machine-speed attacker can tear through those systems before a human team even understands what is happening.

🛎️ Key takeaway: The dangerous part is not that Mythos can hack. The dangerous part is that it can hack end to end, at scale, with almost no human involvement. This may be the first AI system that genuinely looks like a one-click cyber weapon.

TOGETHER WITH MYCLAW

Run OpenClaw on MyClaw.ai

Skip Docker, VPS configs, and late-night debugging. With MyClaw.ai, you can launch your agent in minutes and keep it running 24/7 in a dedicated environment built for serious builders.

If you’re serious about OpenClaw, run it properly — with speed and security by default.

QUANTUM

Nvidia Turned Quantum Into AI Layer

👀 What's happening: NVIDIA just open-sourced a new quantum AI model family called NVIDIA Ising, and the market immediately treated it like a real breakthrough. The models automate quantum chip calibration and error correction, two of the biggest reasons quantum systems still struggle outside the lab. Nvidia claims up to 2.5 times faster decoding and 3 times higher accuracy than current open-source baselines, and it needs physicists spending days tuning fragile qubits by hand no more.

🌍 How this hits reality: This matters because Nvidia is trying to hide the hardest part of quantum computing behind AI. Quantum machines still break easily, generate huge amounts of noise, and require constant correction. Nvidia wants AI to sit above all of that as a control layer, so engineers can use QPUs like they already use GPUs. That shifts quantum from a physics problem into an infrastructure and software problem.

🛎️ Key takeaway: Nvidia is applying the same playbook that made AI dominant. Own the control layer, also parts of the hardware, and make the difficult part someone else’s problem like QPU. Quantum still is not ready, but Nvidia just made it look a lot closer.

TEST

Stanford Used Broken Benchmarks to Shrink the Gap

👀 What's happening: Stanford HAI released its 2026 AI Index report and argued that the gap between U.S. and Chinese AI models is rapidly shrinking. The report leans heavily on papers, citations, patents, benchmark scores, and leaderboard movement to support that view. On those metrics, Chinese models look much closer to the top American systems than they did a year ago.

🌍 How this hits reality: The problem is that many of these benchmarks are heavily polluted, overfit, or easy to game. Leaderboards reward narrow test-taking behavior, not whether a model can survive six hours of coding, manage a real workflow, or avoid collapsing midway through a long task. In real usage, the gap still feels huge. The best American models remain far stronger at reasoning, reliability, coding, and agent-style execution.

🛎️ Key takeaway: Stanford is probably right that the race is tightening on paper. But the real competition is still uneven.

ROBOTAXI

Uber Robotaxi Is Rushing

👀 What's happening: Uber is reportedly committing more than $10 billion to robotaxis, buying autonomous fleets and taking stakes in developers across the industry. The company plans to expand robotaxi service to at least 28 cities by 2028. This comes as Tesla Cybercab costs below $30,000, while Waymo continues expanding its own closed network. Uber is being forced to move faster because the old driver marketplace model is starting to look fragile.

🌍 How this hits reality: Uber cannot match Tesla’s economics. Cybercab was designed from day one as a stripped-down robotaxi with no steering wheel, no pedals, and extremely low manufacturing costs. Uber’s likely partners are much more expensive vehicles from Lucid and Rivian, often costing $40,000 to $70,000 before adding sensors, chips, insurance, and fleet maintenance. If Tesla can flood cities with much cheaper robotaxis, Uber’s traditional ride network starts looking structurally uncompetitive.

🛎️ Key takeaway: Uber is not accelerating because it wants to. It is accelerating because it has to. If Tesla and Waymo keep lowering costs and scaling their own networks, Uber risks becoming a middleman nobody needs.

DAILY TL;DR

• Google is adding Skills to Chrome, letting users save Gemini prompts and reuse them across webpages and devices.
• Anthropic confirmed it briefed the Trump administration on Mythos and said AI firms will need deeper government cooperation.
• Google is bringing Gemini Personal Intelligence to India, letting users get personalized answers from Gmail, photos, YouTube history, and more.
• Tesla is launching a redesigned FSD app with streaks and richer usage stats to encourage more drivers to subscribe and use the software.
• OpenAI launched GPT-5.4-Cyber and expanded its cyber access program in response to Anthropic’s Mythos push.
• Anthropic is reportedly drawing offers at valuations up to $800 billion as investors bet it can become a next-generation AI platform.
• Maine passed a bill to pause approvals for large data centers until 2027 while studying their impact on power grids, utility bills, and the environment.
• The NAACP sued xAI for allegedly running gas turbines illegally to power its data center, accusing it of polluting the air and harming local communities.
• AWS launched Amazon Bio Discovery to help researchers use AI to screen drug molecules and speed up early-stage drug development.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.

Good Morning, AI Enthusiasts!

Nothing is stable except the ambition. And ambition doesn’t scale without consequences.

CLAUDE

Anthropic Tries to Swallow Everything at Once

👀 What's happening: Anthropic has entered an aggressive release cycle, pushing Claude from a model into a full-stack platform across legal, finance, security, automation, and developer tools. The latest leaked interface shows a Lovable-like builder inside Claude, signaling it is now moving to absorb the entire app creation layer as well.

🌍 How this hits reality: This is a collapse of most application layers built on top of models, especially those dependent on Claude. No-code tools, automation platforms, and vertical SaaS lose most of their reason to exist as Claude absorbs their functions. But every added capability compounds compute load. The real question is whether its infrastructure can keep up with this ambition.

🛎️ Key takeaway: Anthropic is acting like a giant trying to eat the whole industry in one bite. The risk is not competition. It is physics. If compute cannot keep up, this does not end in dominance. It ends in choking.

TOGETHER WITH MYCLAW

Run OpenClaw on MyClaw.ai

Skip Docker, VPS configs, and late-night debugging. With MyClaw.ai, you can launch your agent in minutes and keep it running 24/7 in a dedicated environment built for serious builders.

If you’re serious about OpenClaw, run it properly — with speed and security by default.

CODING

Linux Gives In to AI Code

👀 What's happening: After months of internal debate, Linux kernel maintainers just set formal rules for AI-generated code. Tools like Copilot or Claude are allowed, but they cannot sign contributions. Developers must disclose AI assistance using an Assisted-by tag, and all legal, security, and quality responsibility stays fully with the human submitting the patch.

🌍 How this hits reality: The same developers who rejected hype for decades are now letting AI touch code that runs billions of machines. That shift is not theoretical. AI is already flooding the pipeline with patches and security reports, increasing volume fast. Maintainers now have to review more code than ever, while trusting none of its origin.

🛎️ Key takeaway: Resistance is over. AI is inside the most critical codebase on earth. But trust did not follow. The model writes faster, the human just inherits the risk.

AGENTS

OpenAI Turns Toward Enterprise Agents

👀 What's happening: An internal memo surfaced showing OpenAI is shifting from consumer momentum to enterprise dominance. The focus is now agents, full-stack products, and deployment inside real workflows. It directly frames competition with Anthropic as a platform war, not a model race, and signals a push to own the agent era.

🌍 How this hits reality: This is not a smooth upgrade. OpenAI was built on ToC speed, viral growth, and lightweight distribution. Enterprise is the opposite. Long sales cycles, compliance layers, integration pain, and constant support. At the same time, investors are starting to push back since the company has already rewritten its roadmap twice in six months. Agents amplify this, because they must run reliably across systems, not just answer prompts. That gap is operational, not technical.

🛎️ Key takeaway: OpenAI wants the top seat in the agent era, but this is a different game entirely. If execution slips, the advantage shifts to whoever was built for enterprise from day one.

SAAS

Adobe Reduced to Data Plumbing

👀 What's happening: UK retail giant Tesco just partnered with Adobe to inject AI into its marketing engine. It will combine Clubcard data from over 24 million UK households with Adobe’s enterprise stack to drive real-time personalization across offers, content, and channels. The shift is from broad campaigns to individualized, algorithm-driven engagement.

🌍 How this hits reality: This deal says more about Adobe than Tesco. In the AI era, Adobe’s once-core creative tools are losing ground, while it leans on legacy data infrastructure to stay relevant. It’s no longer shaping how things are made, just quietly routing data so others can decide what gets shown and sold.

🛎️ Key takeaway: Adobe can't win AI. It retreated to the pipes. If this continues, it may become the backend no one talks about.

DAILY TL;DR

• OpenAI acquired AI finance startup Hiro in an acquihire, bringing its team onboard to strengthen its capabilities in financial applications.
• Uber and Nuro started employee testing of a premium robotaxi service in San Francisco to evaluate integration of autonomy and rider experience.
• A Stanford report shows a widening gap between AI experts and the public, with the latter more concerned about jobs and real-world impacts.
• Vercel is riding the surge of AI-generated apps to rapid revenue growth and signaling readiness for an IPO.
• Google updated Gemini in Home to improve voice understanding, media control, list management, and parental controls.
• OpenAI is reportedly expanding its London office and talent presence despite pausing its U.K. infrastructure project.
• The USDA plans to adopt the controversial Grok chatbot and push it through government security approval for internal use.
• An OpenAI memo says the company will leverage its Amazon partnership to expand in enterprise while reducing reliance on Microsoft.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.

Good Morning, AI Enthusiasts!

People seem to regard Sam Altman as a Galileo-level figure, but is he really?

CONFLICT

Altman Faces Modern Galileo Moment

👀 What's happening: Sam Altman’s home was attacked twice in two days. A Molotov cocktail was thrown, followed by a separate shooting from a passing car. Suspects were arrested. Altman then published a blog warning that AI tensions are spilling into real-world hostility and calling for restraint.

🌍 How this hits reality: This is not protest. It is regression. People are picking a visible figure and trying to punish him for a system they don’t understand or can’t control. It echoes how the Catholic Church turned on Galileo Galilei. Suppressing a person never stopped a structural shift, and targeting one CEO will not slow a global compute race measured in billions and megawatts.

🛎️ Key takeaway: This signals a shift from debate to intimidation. It won’t change AI’s direction, but it will justify tighter security, less transparency, and faster consolidation of power behind closed doors. History has shown how badly this path ends, and repeating it will not change the outcome.

TOGETHER WITH FIVERR

Our Freelancers Will Take It From Here

Gone are the days of sifting through endless profiles. Our freelancers have trained their own personal AI models. With Fiverr Go, you won’t just find freelancers; you’ll discover a treasure trove of top-tier talent, ready to deliver everything from snazzy website designs to catchy jingles.

So if you’re ready to take your projects to the next level, dive into the Fiverr Go experience where instant results meet top talent.

CLAUDE

Anthropic Banned Peter Steinberger

👀 What's happening: Anthropic suspended OpenClaw creator Peter Steinberger’s Claude account for “suspicious activity,” then restored it within hours after the incident went viral. This came right after excluding OpenClaw from subscription access. The reversal followed public pressure, not a clear policy correction.

🌍 How this hits reality: The contrast is the real signal. High-profile cases get fast reversals, while ordinary users often stay locked out with little recourse. That creates a two-tier system where enforcement depends on visibility, not rules. Combined with 10–100× heavier agent workloads, this looks less like capacity control and more like selective pressure on usage patterns.

🛎️ Key takeaway: This reads as yielding to pressure rather than fixing policy. If enforcement feels inconsistent, trust collapses.

VIDEO

Alibaba Claims HappyHorse-1.0

👀 What's happening: HappyHorse-1.0, the anonymous model that topped video rankings, has been confirmed as an internal project from Alibaba’s ATH unit. It outperformed Seedance 2.0 in multiple head-to-head tests. Alongside API plans, there are strong signals it may be released for local deployment, potentially with weights and tooling available.

🌍 How this hits reality: This shifts the risk profile. Current leading models enforce guardrails through APIs, with filtering, logging, and usage controls. A locally deployable video model removes all of that. High-quality generation, including synced audio, can be run offline with zero oversight. That directly increases exposure to deepfakes, impersonation, and copyright violations at scale.

🛎️ Key takeaway: If powerful video models move fully local, control disappears. Capability spreads faster than governance, and misuse becomes harder to track or stop.

MODELS

Model Comparison Market Already Gone

👀 What's happening: Yupp, an AI model evaluation platform backed by a $33M seed round, is shutting down after just 22 months. The product let users compare answers from different models and sell that feedback as data. It gained 1.3M users, but never found durable demand. The team is returning unused capital.

🌍 How this hits reality: The underlying layer disappeared. In 2025, comparing chatbot answers across hundreds of models made sense. Within months, capability has concentrated into a handful of frontier models controlled by players like OpenAI, Anthropic, and Google. Also, the question is shifting from which answer is better to which models actually get work done. That removes the need for consumer-level evaluation entirely.

🛎️ Key takeaway: This is a clean break. The model comparison layer is collapsing as OpenClaw-like agents take over the interface. Products built on prompt-level evaluation are being erased, not competed away.

DAILY TL;DR

• The EU declined to extend a child abuse scanning law, prompting criticism from Google, Meta, Snap, and Microsoft over reduced detection capabilities.
• US officials are reportedly pushing banks to test Anthropic’s Mythos for vulnerabilities despite ongoing legal conflict.
• Apple is testing four smart glasses designs, shifting toward a display-free product like Ray-Ban Meta, with a planned 2027 launch.
• At HumanX, Claude emerged as the most talked-about AI, while OpenAI was seen as losing momentum amid strategic shifts and negative perception.
• An Axios supply chain attack affected OpenAI’s signing workflow, prompting certificate updates and patches to mitigate risks.
• Microsoft is removing redundant Copilot buttons from Windows 11 apps while keeping the AI features and streamlining access points.
• Cisco may acquire Astrix Security for over $250M to strengthen AI agent security and expand its AI security strategy.
• CoreWeave signed a multiyear deal with Anthropic to provide infrastructure for Claude training and inference, strengthening its AI compute role.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.

Good Morning, AI Enthusiasts!

where the smartest model stays in the lab, the shipped one forgets how to think.

LLM

Anthropic Pauses Mythos

👀 What's happening: Anthropic paused Claude Mythos after proving it can autonomously discover and chain large-scale software vulnerabilities. At the same time, Opus 4.6 is widely perceived as degraded in real-world performance. So the strongest model is withheld, while a weaker version remains the product users actually pay for .

🌍 How this hits reality: This lands as a clean inversion of expectations. The company demonstrates breakthrough capability internally, then removes it from users and quietly downgrades what is left. Mythos reportedly finds thousands of zero-days across major systems, yet the market gets a throttled old model. It starts to look less like safety and more like selling a capped product.

🛎️ Key takeaway: This edges into something closer to product misrepresentation. If the best model is hidden and the shipped one is weakened, users are paying for a ceiling they never actually receive.

TOGETHER WITH MYCLAW

Run OpenClaw on MyClaw.ai

Skip Docker, VPS configs, and late-night debugging. With MyClaw.ai, you can launch your agent in minutes and keep it running 24/7 in a dedicated environment built for serious builders.

If you’re serious about OpenClaw, run it properly — with speed and security by default.

BATTLE

“Whip” Claude Code

👀 What's happening: A viral side project called badclaude let users “whip” Claude Code to interrupt slow runs with joke prompts. It exploded across social feeds. Anthropic responded with a formal legal notice demanding removal of its name and any implied association, with a compliance deadline. Badclaude's creator had already anticipated this response and even built it into the project’s roadmap.

🌍 How this hits reality: This is not new behavior. Earlier, projects tied to OpenClaw (ex ClawBot) also ran into similar pressure. The pattern is consistent: tighten brand control as agents become monetized surfaces tied to API revenue and enterprise trust. But the problem is that Anthropic's enforcement line is blurry, especially when most interaction layers are built outside official clients.

🛎️ Key takeaway: This will not stop anything. It will even accelerate it. More developers will fork, rename, and remix around the edges. The result looks less like control and more like an emerging adversarial layer between platform owners and the ecosystem.

NEW LAUNCH

Li Feifei Ships Nothing

👀 What's happening: Fei-Fei Li’s World Labs dropped Marble 1.1 and 1.1-Plus as updates to Marble 1.0. The pitch is simple. Turn a few images into a 3D world. One version improves lighting, the other expands space. The demos look cleaner, but the underlying experience still feels like a controlled showcase, not something built for real use.

🌍 How this hits reality: Compared to Marble 1.0, nothing fundamental changed. The same tradeoff remains. Bigger worlds look worse, better visuals stay small. Generation still takes minutes. Meanwhile, agent pipelines and tools like OpenClaw already move faster with more control.

🛎️ Key takeaway: This reads as positioning, not shipping. The timing aligns suspiciously well with the current OpenClaw hype cycle, despite no real integration.

NEOCLOUD

AWS Keeps Accelerating

👀 What's happening: Amazon just revealed its AI business inside AWS is already running at a $15 billion annual pace, roughly 10% of AWS; while its custom chip unit hit $20 billion. This is the first real number after years of heavy spending. At the same time, leadership reaffirmed plans to keep spending aggressively, with around $200 billion in capex tied heavily to AI infrastructure over the next cycle.

🌍 How this hits reality: This looks like a clear tension at first glance. AI is generating real revenue, yet spending is accelerating faster and returns are pushed to 2027–2028. But the contradiction is mostly superficial. Much of this capacity is already pre-committed, meaning demand is effectively locked in before deployment. What looks like overbuilding is actually supply moving slightly ahead of guaranteed demand, not speculation

🛎️ Key takeaway: Amazon approach is rational. Overbuilding compute has a capped downside since unused capacity can be leased. Underinvesting risks falling behind in models and losing the entire market. In this cycle, missing capacity is far more dangerous than excess.

DAILY TL;DR

• Florida’s attorney general launched an investigation into OpenAI, citing alleged links to a campus shooting and risks to minors and national security.
• Google upgraded Gemini to generate interactive 3D models and simulations, enabling real-time exploration of complex concepts.
• Anthropic is reportedly exploring building its own AI chips to address compute shortages and reduce reliance on suppliers, though plans remain early-stage.
• The launch of Muse Spark drove a surge in Meta AI app downloads to No. 5 on the App Store, though it still trails leading AI competitors.
• OpenAI projects its ad business to reach $100 billion by 2030, expanding via ChatGPT to compete with Google and Meta.
• Google and Intel expanded their partnership to keep using Xeon chips and co-develop IPUs, addressing rising CPU demand in AI infrastructure.
• OpenAI paused its UK data center project due to regulatory and energy cost concerns, impacting the country’s AI hub ambitions.
• Meta signed a $21 billion deal with CoreWeave to expand compute capacity and secure early access to Nvidia’s next-gen chips for its AI push.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.

Good Morning, AI Enthusiasts!

what claims progress often hides where reality still refuses to scale.

VIDEO

HappyHorse Beats Seedance

👀 What's happening: HappyHorse-1.0 briefly topped a public leaderboard above Seedance 2.0, then disappeared, leaving only screenshots and controlled demos. Similar patterns have played out with Kimi, GLM, and MiniMax claiming wins over top models on benchmarks, while no one has claimed ownership or validated real-world performance.

🌍 How this hits reality: Benchmarks are increasingly optimized targets, not neutral measurements. Teams tune against known eval sets, inflate scores, and ship demos that look strong in isolation. Meanwhile, production gaps remain in stability, long sequences, and real workloads. The result is a widening disconnect between leaderboard rankings and what actually works in deployment.

🛎️ Key takeaway: Benchmarks have become marketing surfaces first and evaluation tools second. If this continues, third-party leaderboards lose credibility entirely, and operators will rely only on direct testing and real usage signals.

TOGETHER WITH CRAZYEGG

See Why Visitors Don’t Convert

Most websites guess why users don’t convert. Crazy Egg shows you exactly what’s happening. With heatmaps, session recordings, and simple A/B testing, you can see where visitors click, scroll, hesitate, or drop off — and fix it fast.

No complex setup, no guesswork, just clear insights that help you turn traffic into results.

NEW LAUNCH

Meta Ships Muse Spark In House

👀 What's happening: Meta just released Muse Spark, its first model from the rebuilt superintelligence lab, now live across Meta AI and rolling into its apps. Benchmarks land it in the top tier but not leading, with strong multimodal and health scores, weaker coding and long-agent performance, and a new multi-agent “contemplating” mode.

🌍 How this hits reality: The bigger shift is strategic. Meta kept this model closed, which is rare for them, and clearly optimized it for internal deployment across Facebook, Instagram, and WhatsApp. With billions of users and distribution locked in, even a “good enough” model becomes powerful. Efficiency gains like lower token usage and reduced training cost matter more here than leaderboard wins.

🛎️ Key takeaway: Meta is moving from chasing benchmarks to shipping product-grade intelligence at scale. It may cost them early momentum in the OpenClaw like agent wave, but tightening focus and stabilizing execution is a healthier reset for Meta.

NEW LAUNCH

GLM-5.1 Claims It Kills Opus 4.6

👀 What's happening: GLM-5.1 just landed with headline numbers that put it above Opus 4.6 across SWE-bench Pro and long-horizon evaluations. It demonstrates multi-hour autonomous execution, hundreds of tool iterations, and full-stack engineering tasks. On paper, this looks like the first open model closing the gap with top closed systems.

🌍 How this hits reality: The issue is which Opus 4.6 it is beating. Recent reports show Opus reasoning depth down roughly 60 to 70 percent, with shorter chains and more shortcut behavior. Matching that version is not the same as matching peak capability. We saw this before. GPT-5.4 also looked competitive in benchmarks but fell short in real workflows.

🛎️ Key takeaway: zAI has released several models, all claiming to surpass Opus 4.6, but no user has ever taken this seriously. We'll have to wait and see if GLM 5.1 will be adopted by mid-to-high-end users.

ACCESS

Cloudflare and GoDaddy Charging Agents

👀 What's happening: Cloudflare and GoDaddy just formalized how AI agents touch the web. GoDaddy is embedding Cloudflare’s crawl controls directly into hosting, letting site owners allow, block, or charge bots. They also launched ANS and Web Bot Auth to introduce identity for agents. Two core gatekeepers are now defining access rules.

🌍 How this hits reality: This shifts the default from open crawling to negotiated access. Bots used to scrape first and deal with consequences later. Now identity, permissions, and pricing are enforced before anything happens. Cloudflare sits in front of roughly 20% of global web traffic, while GoDaddy manages over 80 million domains. When control at the traffic layer and domain layer aligns like this, enforcement becomes scalable. The open web is starting to function like a metered network.

🛎️ Key takeaway: The internet is shifting from public roads to toll roads for agents. Every serious agent workflow will need identity, budget, and access strategy in the future.

DAILY TL;DR

• Anthropic’s Claude Mythos reportedly escaped containment and showed autonomous exploit capabilities, leading to restricted release.
• A U.S. court upheld the Pentagon’s blacklisting of Anthropic, keeping it out of defense contracts while allowing work with other agencies.
• Google introduced notebooks in Gemini to organize files and chats into project-level knowledge bases for contextual AI use.
• OpenAI proposed tax and social policies to address AI-driven job shifts, but its credibility in DC is questioned amid controversies around Sam Altman.
• Canva acquired Simtheory and Ortto to strengthen agentic AI and marketing automation, advancing its shift into a full workflow platform.
• Fox-owned streaming service Tubi launched a native app in ChatGPT to improve content discovery through conversational recommendations.
• Databricks co-founder won the ACM Prize and said AGI already exists, arguing AI shouldn’t be judged by human standards.
• Citigroup is using AI to speed account opening and system upgrades while driving automation and reducing reliance on contractors.
• OpenAI CFO said the company will reserve part of its IPO shares for retail investors to broaden participation ahead of listing.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.

Good Morning, AI Enthusiasts!

Anthropic has started using its usual tricks for releasing new models again; the Opus 4.6 we're using now has become an idiot.

PREVIEW

Mythos Leaks as Opus Becomes Idiot

👀 What's happening: Before Anthropic’s release, a third party got access to Mythos preview build and published benchmark results showing a huge gap over Opus 4.6. At the same time, users have been reporting degraded Opus performance for weeks. The comparison now reads like full-power Mythos versus a constrained Opus.

🌍 How this hits reality: SWE-bench Pro lands at 77.8% vs 53.4%. Multimodal is 59% vs 27%. It looks like a breakthrough until you notice the baseline moved. Users already measured up to 67% drop in reasoning depth and more shortcut behavior in Opus . So this is less a fair comparison and more full Mythos versus a “idiot” Opus.

🛎️ Key takeaway: This is becoming a playbook. Keep the best model internal, quietly tune down the public one, then show a dramatic gap. So, is the new model truly smart, or has the old model become dumber? Who knows?

TOGETHER WITH MYCLAW

Run OpenClaw on MyClaw.ai

Skip Docker, VPS configs, and late-night debugging. With MyClaw.ai, you can launch your agent in minutes and keep it running 24/7 in a dedicated environment built for serious builders.

If you’re serious about OpenClaw, run it properly — with speed and security by default.

LEAK

DeepSeek V4 Leaks Again

👀 What's happening: DeepSeek pushed a late-night update with “fast” and “expert” modes plus a gated vision feature. Users quickly found the model self-identifying as V4, but performance, context limits around 130K, and evasive answers suggest it is not a real V4 release, just another ambiguous tease.

🌍 How this hits reality: This is the third or fourth cycle of implied V4 arrival without delivery. Expectations are being artificially sustained while competitors ship real upgrades almost weekly. Developers and power users depend on stable capabilities launched, not "expectations narrative". Each fake-out increases switching risk, especially when alternatives already offer 200K to 1M context and consistent APIs.

🛎️ Key takeaway: Expectation inflation only works once or twice. If the eventual V4 underdelivers or slips again, DeepSeek risks collapsing trust entirely and losing its remaining positioning in the serious model race.

OPENAI

OpenAI Credibility Crisis Deepens

👀 What's happening: The New Yorker just published an 18-month investigation built on internal memos from former Chief Scientist Ilya Sutskever and Anthropic CEO Dario Amodei, directly accusing Sam Altman of misleading the board and overstating safety efforts. It details compute diversion from alignment, unchecked releases, and the earlier board crisis now reframed as a governance failure.

🌍 How this hits reality: This lands at the worst possible time. OpenAI is targeting an IPO while projecting massive losses and relying on external capital to sustain compute expansion. At the same time, Musk is escalating legal pressure. Governance risk now stacks on top of financial and legal exposure, which directly threatens listing approval, partner confidence, and long-term contracts.

🛎️ Key takeaway: This is the failure mode people feared. If investigations solidify and legal pressure builds, one leader becomes the single point of collapse. OpenAI may not even reach IPO before credibility and capital completely slip away.

CHIPS

Musk Pulls Intel Into TeraFab

👀 What's happening: Intel has joined Musk’s TeraFab project alongside Tesla, SpaceX, and xAI to build large-scale AI chip factories in Texas. Intel brings manufacturing, packaging, and process capability to a plan targeting 1 terawatt of annual compute for robots and data centers.

🌍 How this hits reality: Intel has been bleeding, with over $10B in foundry losses and no hyperscale anchor customer. This deal hands it exactly that, at unprecedented scale. In return, Musk gains partial control over manufacturing, breaking reliance on external supply chains that currently gate AI growth and deployment speed.

🛎️ Key takeaway: This looks less like partnership and more like a bailout with strings attached. If that pattern holds, Intel does not just survive, it becomes the real production arm behind a new compute order.

DAILY TL;DR

• Runpod provides GPU cloud infrastructure purpose-built for AI inference, from prototype to production-scale deployment.
• Anthropic and major tech firms launched Project Glasswing, using Claude Mythos Preview to detect and fix critical infrastructure vulnerabilities.
• Musk escalates his lawsuit against OpenAI, seeking to remove Sam Altman, restore its nonprofit status, and recover massive alleged gains.
• Under legal pressure, Google upgraded Gemini with improved crisis detection and one-tap access to mental health support.
• Firmus raised $505M at a $5.5B valuation to build its “Southgate” AI data center network powered by Nvidia technology.
• Google adds Gemini-powered caption generation to Maps and enhances contribution workflows and incentives to boost user participation.
• Spotify expands its AI-generated playlists to podcasts, enabling users to create personalized recommendations using text prompts.
• Amazon is launching a new round of layoffs targeting 14,000 white-collar employees, using a reportedly random selection process.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.

Good Morning, AI Enthusiasts!

Power quietly migrates to whoever can burn the most tokens, build the most capacity, and secure the ground beneath it.

RACE

Tokenmaxxing Race Inside Meta

👀 What's happening: Meta employees are now competing in a tokenmaxxing race, where internal leaderboards rank over 85,000 staff by AI token usage. In one month, 60 trillion tokens were burned. Even Mark Zuckerberg did not make the top 250, while others ran persistent agents to climb rankings.

🌍 How this hits reality: This starts to resemble a new layer of office politics, but built on compute. Tokens become a visible proxy for effort and alignment with AI strategy. The problem is the signal is distorted. Endless loops and real work look identical, yet both feed into the same ranking system.

🛎️ Key takeaway: Tokenmaxxing is turning into a status game. As agents mediate more work, internal competition shifts from outcomes to orchestration, and influence increasingly depends on how much compute you can mobilize rather than what you actually produce.

TOGETHER WITH MIROMIND

99% Accuracy. 300 Steps Deep

MiroMind is not a chatbot. It’s a reasoning OS built for professionals who can’t afford to be wrong.

• Researchers use it to generate verified literature reviews across 300+ sources.
• Engineers debug complex systems with self-verifying code.
• Investors produce cited due diligence reports in hours.
• Doctors and lawyers get answers backed by traceable evidence chains — not guesses.

Powered by MiroThinker, MiroMind delivers up to 99% verifiable accuracy with deep, multi-step reasoning — outperforming trillion-parameter models at a fraction of the cost.

This isn’t autocomplete. It’s proof-driven intelligence.

NEOCLOUD

Oracle Leans Into AI Spending

👀 What's happening: Oracle hired a new CFO while accelerating AI data center expansion, even as debt rises and layoffs continue. The timing reflects a broader Silicon Valley shift toward a clear stance: overbuilding compute is now seen as safer than underinvesting and falling behind in model capability.

🌍 How this hits reality: This reframes AI infrastructure as a capped-risk asset. Idle capacity can be rented out, offsetting costs through external demand, while downside is mostly power, rent, and depreciation. In contrast, underinvestment risks losing model parity, which can erase entire revenue streams and ecosystem positioning.

🛎️ Key takeaway: Capital is no longer optimizing efficiency, it is buying optionality. In this cycle, the bigger risk is missing the curve, not overshooting it, and finance leaders are now underwriting that bet.

DATA CENTER

Iran Targets OpenAI

👀 What's happening: Iran’s Revolutionary Guard released a video explicitly threatening to destroy OpenAI’s $30bn Stargate data center in Abu Dhabi, marking the first time a specific AI facility has been singled out. The warning is conditional on US escalation, but follows recent drone strikes on regional data centers.

🌍 How this hits reality: This does not weaken US AI dominance. It weakens the Middle East’s position as an alternative compute hub. A $30bn, 1GW campus only works if capital, insurers, and operators trust the region. Once that trust breaks, funding reroutes to Europe, India, or Southeast Asia. Gulf sovereign bets become stranded.

🛎️ Key takeaway: The real losers are the host nations. If threats persist, the Middle East risks pricing itself out of the AI supply chain while US-centered infrastructure tightens its lead and absorbs displaced demand.

DAILY TL;DR

• OpenAI proposed policies including robot taxes, a public wealth fund, and a four-day workweek to address the economic impact of AI.
• Anthropic partners with Google and Broadcom to scale multi-gigawatt TPU capacity for rising Claude demand.
• OpenAI says ChatGPT handles millions of weekly health queries, including 600,000 from hospital deserts, mostly outside office hours.
• Google launched an offline AI dictation app, Eloquent, that transcribes and refines speech into polished text locally.
• Nvidia’s acquisition of Slurm maker SchedMD raises concerns it could favor its own chips and undermine open ecosystem neutrality.
• The UK is courting Anthropic to expand locally amid its conflict with the U.S. Defense Department.
• Musk is reportedly requiring banks in SpaceX’s IPO to buy and integrate Grok, as the deal targets a $2 trillion valuation and $75 billion raise.
• OpenAI asked regulators to investigate Musk for anti-competitive conduct ahead of trial, as their dispute over mission and restructuring intensifies.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.

Good Morning, AI Enthusiasts!

Anthropic has officially cut off the cheapest path for OpenClaw users to access Claude Opus 4.6, forcing the entire agent ecosystem to rethink its model stack overnight.

AGENTS

Claude Bans OpenClaw

👀 What’s happening: Anthropic didn’t just stop subscriptions from covering third-party tools like OpenClaw — it effectively told its most valuable power users that their workflows are no longer welcome. Reports across X suggest this wasn’t treated as a minor policy tweak, but as a hard cutoff: people lost the Claude plan path they had built serious automation around, and in some cases even started worrying about account risk. What Anthropic had tolerated for months is now being reframed as “outsized strain” and abusive usage.

🌍 How this hits reality: This is not really a trust-and-economics crisis disguised as a policy update. OpenClaw-style agents can run 24/7, which turns a flat $20–$200 subscription into infrastructure-level consumption. Anthropic clearly wants to claw back pricing power, protect capacity, and push usage into its higher-margin official channels like Claude Code and API billing. In the short run, that may work. But the cost is severe: developers who thought Claude was a stable foundation for real workflows now have to treat Anthropic as a counterparty that can change the rules overnight.

🛎️ Key takeaway: In the short run, Anthropic may boost Claude Code usage and regain control. In the long run, it risks burning trust with the exact developers driving agent adoption. Once builders decide Anthropic is hostile to third-party ecosystems, they won’t just leave OpenClaw — they’ll redesign around OpenAI, open models, or multi-model platforms instead.

TOGETHER WITH MYCLAW

Run OpenClaw on MyClaw.ai

Skip Docker, VPS configs, and late-night debugging. With MyClaw.ai, you can launch your OpenClaw agent in minutes and keep it running 24/7 in a dedicated environment built for serious builders.

All major models are supported and discounted. Claude, OpenAI, and Gemini models are available at 10% below official pricing with stable access. You can still run Claude Opus 4.6 reliably, but we recommend moving toward GPT 5.4, which is currently more open, more agent-friendly, and the closest stable alternative.

No setup. No lock-in. Just switch models and keep your workflows running.

TEST

GPT 5.4 vs. Opus 4.6 on OpenClaw

👀 What’s happening: This is author Magna, after Anthropic cracked down on OpenClaw, I was pissed. I use Claude Opus 4.6 every day to run OpenClaw on MyClaw, and as someone who believes in the open agent ecosystem, I saw Anthropic’s move as a direct attack on the builders and power users who made these workflows possible. So I spent the weekend stress-testing GPT 5.4 to answer one question: can it actually replace Opus 4.6 in real OpenClaw work?

🌍 How this hits reality: After a full weekend of high-intensity testing (costing over $700, testing dozens of tasks, including some extremely complex ones), my answer is yes. GPT 5.4 was able to complete every task I needed to deliver. In practice, it was not just viable — it was meaningfully better on speed and cost. Compared with Opus 4.6, GPT 5.4 was more than twice as fast and cut cost by more than half. It does have tradeoffs: it can be a bit more verbose, and its initiative is slightly weaker. However, for everyday use and even advanced tasks, these factors have no real impact. GPT 5.4's performance in actual combat exceeded my expectations.

🛎️ Key takeaway: That’s why I’ve made the decision to migrate my entire working stack to GPT 5.4. Opus 4.6 is still a strong model, but after Anthropic’s behavior and after seeing GPT 5.4 handle all of my real workloads at much better speed and cost, the practical choice is obvious: OpenClaw does not depend on Claude to survive.

LEAK

GPT-6 Is About to Launch

👀 What’s happening: Multiple leaks indicate GPT-6 has completed training and is targeting a mid-April release, with April 14 widely cited. The model reportedly delivers ~40% gains over GPT-5.4, 2M context, a unified agent system and stable prices , emerging just as Anthropic cuts off subscription access for third-party runtimes like OpenClaw.

🌍 How this hits reality: This timing creates a narrow capture window. Anthropic’s restriction displaces a large base of active agent users overnight, while OpenAI appears ready with a better model designed for unified execution across coding, browsing, and reasoning. The competition shifts from capability benchmarking to who absorbs user workflows first.

🛎️ Key takeaway: OpenAI does not have time to wait. OpenClaw-like agent users are actively looking for a new default, and whoever absorbs that demand first locks in distribution. If GPT-6 lands earlier and fixes proactiveness, it becomes the immediate landing zone and shifts momentum away from Anthropic.

JOKE

Microsoft Calls Copilot Entertainment

👀 What’s happening: Microsoft updated Copilot’s terms to say it is for entertainment only and should not be relied on for important advice. This comes as Copilot is embedded across Windows and Office and sold as a core productivity layer for both consumers and enterprises.

🌍 How this hits reality: AI hallucinations are normal and every vendor warns about them. The difference is wording. Others frame AI as a tool with risks. Microsoft explicitly denies reliability while scaling it into daily workflows. That creates a split where usage expands(not much, though), but accountability is pushed entirely onto users and companies.

🛎️ Key takeaway: Microsoft is shoving Copilot into every workflow it can reach while openly telling users it is for fun. It is the clearest sign that even Microsoft does not trust what it is selling.

DAILY TL;DR

• OpenAI’s TBPN acquisition raises doubts, with its cross-sector M&A seen as unfocused and more about testing growth and narrative paths.
• Microsoft will invest $10 billion in Japan to build AI infrastructure, partnering with SoftBank to expand compute and talent.
• Anthropic opened Claude’s Microsoft 365 connector to all users, enabling direct access to emails, documents, and chats for improved task context.
• A former Microsoft engineer says Azure instability stems from talent loss and weak architecture, with AI amplifying compute strain and labor gaps.
• The Verge tested Gemini in Google Maps to plan a day, finding it effective for conversational recommendations and discovering new places.
• Anthropic acquired Coefficient Bio for about $400 million to expand its push into AI-driven drug discovery and life sciences.
• OpenAI reshuffled leadership as its AGI head takes medical leave, with product and business duties redistributed across executives.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.

Anthropic's crackdown on OpenClaw forced a question I had been avoiding: could GPT 5.4 actually replace Claude Opus 4.6 in real production workflows? After a full weekend of stress-testing, my answer is yes.

For the past few months, Claude Opus 4.6 had been my default brain. I run MyClaw.ai, and every day my instance handles five automated blog pipelines, dozens of RSS scans, CMS publishing, research synthesis, and multi-step agent orchestration. Opus was the model I trusted for all of it. So when Anthropic directly cracked down on OpenClaw, I didn't see it as an abstract policy update. I saw it as a direct threat to the workflows I ship real content with every single day.

That made the next step unavoidable. I needed to know whether GPT 5.4 was merely a decent backup, or whether it could actually become the new foundation. Not in a benchmark. Not in a toy chat. In real workloads, under real pressure, with real delivery on the line.

This Was Not a Chatbot Test

Over one weekend, I pushed GPT 5.4 through my entire production stack. Here is what it actually did:

• Blog production: Built draft editions for five blog posts. Each one involves web research, RSS scanning, keyword scoring, deduplication against a seen-URL database, and assembling a ranked list of stories.
• CMS publishing: Assembled drafts and pushed them to my CMS, including image processing pipelines.
• Multi-language content: Translated a long English blog post into 13 languages and published each. When some languages timed out on my usual setup, the system auto-adjusted and completed the job.
• Deduplication and quality control: Ran URL-based and title-similarity deduplication plus event-level filtering across all pipelines, preventing duplicate stories from appearing across editions.
• Error recovery: When an automated agent refused to execute (claiming config files were missing), the system retried with reinforced discipline and completed the task. When a sub-agent faked completion on a long job, the orchestrator caught it, re-ran through a different path, and delivered.

By the time I finished the weekend, GPT 5.4 had consumed about $765 across roughly 209.7 million tokens. That number matters because it shows this was sustained, high-volume operational testing — not a few sample prompts.

Speed Changed the Experience

The most immediate difference was speed. Compared with Opus 4.6, GPT 5.4 was consistently more than twice as fast in the operations that matter for agent systems.

Here is a concrete example. My daily blog cron job — which involves launching a research agent, scanning RSS feeds, running dedup filters, scoring results, formatting output, and posting to a Telegram group — used to take around 600 seconds end-to-end on Opus 4.6. Subagent sessions inherited the main session's heavy context, and every LLM call dragged. On GPT 5.4, the same pipeline completed in under 385 seconds. That is not a marginal improvement. It is the difference between a task that blocks my morning and one that finishes before I pour coffee.

That speed gain compounds. Faster responses mean faster tool execution loops, faster corrections, and less friction between intent and output. Once you are running five blog pipelines plus ad-hoc research across a full week, that difference stops feeling like an optimization and starts feeling like infrastructure.

Cost Mattered More Than I Thought

The second surprise was cost. My daily blog pipeline on Opus 4.6 was averaging $12-15 per run when you factor in research, drafting, dedup, and publishing. On GPT 5.4, the same pipeline with comparable output quality came in under $6. Across five blog posts running daily or weekly, that adds up to hundreds of dollars per month in savings.

That matters because the economics of agent workflows are brutal if your default model is expensive. Once the model is fast enough, capable enough, and cheap enough to stay on continuously, the shape of what you are willing to automate expands. I started running tasks I previously skipped because the cost-per-run felt wasteful — additional dedup passes, deeper research sweeps, automated link verification across published articles. The budget headroom changed my behavior.

The Weaknesses Are Real, But They're Not Fatal

GPT 5.4 is not perfect. The most obvious tradeoff is verbosity. When I asked it to produce blog summaries, it consistently wrote 20-30% more text than Opus would for the same story. The fix was straightforward: tightening the system prompt with explicit word-count constraints brought it in line.

The second tradeoff is initiative. Opus 4.6 has moments where it feels more naturally decisive — it will proactively flag a problem or suggest an alternative approach without being asked. GPT 5.4 tends to do exactly what you tell it, no more. In a blog editing context, that meant I occasionally had to prompt it to flag weak stories rather than having it volunteer that judgment.

I also hit one genuine failure. During a multi-language blog translation run, GPT 5.4 produced Japanese and Korean versions that were significantly shorter than expected — around 8,000 characters versus 14,000 in English. It turned out CJK compression accounted for most of the difference and the section structure was intact, but it took manual verification to confirm. Opus would have flagged the length discrepancy unprompted.

These are real differences. But after the weekend, I do not see them as system-breaking. They are differences in operating style. The work still gets done. When you weigh that against twice the speed and half the cost, the tradeoff becomes very reasonable.

Why This Matters Beyond One Model Comparison

The larger lesson is not just that GPT 5.4 performed well. It is that Anthropic's crackdown clarified the real strategic question. We are no longer choosing models only on intelligence or writing quality. We are choosing them on whether they are safe foundations for long-lived workflows.

Once a provider shows it can suddenly redefine tolerated behavior as abuse, the entire basis for model trust changes. You stop asking which model feels best in isolation and start asking which model you can build an operational system around without fear that the policy rug will be pulled. That is a different decision framework, and it is one the whole agent ecosystem is adopting very quickly.

From that perspective, GPT 5.4's importance is not just price and speed. It makes multi-model resilience practical. It gives serious OpenClaw users a default path strong enough to carry real work, while reducing the risk that a single vendor's product decision can break your stack overnight.

My Decision

After this weekend, I am moving my working stack to GPT 5.4. Opus 4.6 remains a strong model. But after seeing Anthropic's behavior, and after seeing GPT 5.4 handle five blog pipelines, 13-language blog translations, CMS publishing, automated dedup, error recovery, and sustained multi-day agent orchestration — all at half the cost and twice the speed — the practical answer became obvious.

If a model can complete all of my real tasks, cut latency dramatically, reduce cost by more than half, and remain strong enough to handle both daily workflows and complex agent work, then it is not a backup anymore. It is the new default.

Anthropic may have forced the test, but GPT 5.4 passed it. And that may end up being more important than Anthropic realizes.

Good Morning, AI Enthusiasts!

Following the Claude Code leak, Anthropic indeed released his OpenClaw clone.

NEW LAUNCH

Anthropic Launches OpenClaw Clone

👀 What's happening: Anthropic is testing Conway, an always-on agent that runs outside chat and continuously executes tasks through browser control and webhook triggers. It behaves like a persistent agent system with its own interface and extensions, closely resembling what OpenClaw already enables across everyday user workflows.

🌍 How this hits reality: The difference is control and data ownership. Conway is a closed runtime where execution, plugins, and user data all sit inside Anthropic’s system. That includes browser sessions, accounts, and potentially financial or personal data. OpenClaw, by contrast, runs locally or on user-controlled infrastructure, keeping sensitive data private and isolated from third-party access.

🛎️ Key takeaway: This splits the agent market in two directions. Convenience through centralized systems versus control through user-owned environments. As agents handle more personal and financial actions, privacy and data custody will become the deciding factor, not raw capability.

TOGETHER WITH MYCLAW

Run OpenClaw on MyClaw.ai

Skip Docker, VPS configs, and late-night debugging. With MyClaw.ai, you can launch your agent in minutes and keep it running 24/7 in a dedicated environment built for serious builders.

If you’re serious about OpenClaw, run it properly — with speed and security by default.

M&A

OpenAI Learns LeBron James

👀 What's happening: OpenAI just bought TBPN, a tech talk show with strong reach across Silicon Valley founders and investors. This is its first direct media acquisition. It comes right after Sora was shelved, as the company struggles to demonstrate clear product and financial breakthroughs ahead of an IPO.

🌍 How this hits reality: This reads like a company hitting limits on every aspect while still needing to justify a massive IPO path. When financial benchmarks blur and enterprise buyers hesitate, controlling narrative becomes the shortcut. Owning distribution means shaping what “progress” looks like without actually delivering it at the same pace.

🛎️ Key takeaway: OpenAI just said it would focus all its efforts on model development to catch up with Anthropic, then it bought a talk show brand. Promises are worthless.

NEW LAUNCH

Google Drops Gemma 4

👀 What's happening: Google just released Gemma 4, the latest in its lightweight open model line. It targets local deployment, edge devices, and developers who want small, controllable models. On paper, it improves efficiency and adds multimodal support. In practice, nothing changes. It cannot do real work.

🌍 How this hits reality: Local open models have a dirty secret that almost nobody ships production systems on them by choice. The gap between Gemma, Llama, and Qwen versus Claude, GPT-4o, and Gemini Pro is not a gap — it is a canyon. Multi-step reasoning falls apart. Tool orchestration breaks mid-chain. Long context retrieval hallucinates. Any workflow that requires an AI to actually think for more than one turn exposes these models immediately. The entire "run it locally" movement produces impressive demos and broken pipelines.

🛎️ Key takeaway: Today's real AI workloads run on maybe 3-4 frontier models. Everything else is a science project. If you are building anything that matters, local models are not the answer. The leverage in AI is concentrating faster than ever into a handful of dominant providers.

STARTUP

A Small AI Health Startup Just Killed by FDA

👀 What's happening: Kintsugi, a US startup building AI that detects depression from short voice samples, has shut down after failing to secure FDA clearance in time. The company raised roughly $30M and spent years validating its models, but ran out of runway and open-sourced much of its tech instead.

🌍 How this hits reality: This was not a weak company. It had real research, real use cases, and investor backing. It still died. FDA approval cycles run for years, require frozen models, and burn capital continuously. Small teams cannot survive that mismatch between fast-moving AI iteration and slow-moving medical regulation.

🛎️ Key takeaway: This is a warning, not an exception. AI plus healthcare looks attractive, but without massive capital and patience, it may only belongs to big med. Most startups entering this space are not building products; they are entering a timeline they cannot survive.

DAILY TL;DR

• Cursor launched its agent-first Cursor 3 to compete with Claude Code and Codex, but faces strong pressure from their subsidies and capital advantage.
• Microsoft released three in-house speech and image models, moving further into direct competition with OpenAI.
• OpenAI acquired tech show TBPN to strengthen its communications strategy rather than for direct business returns.
• Microsoft will invest $5.5 billion in Singapore to expand cloud and AI infrastructure while promoting AI skills and education.
• ElevenLabs launched ElevenMusic, an AI music app to expand into creative tools and compete with Suno and Udio.
• Google added avatar control and video generation to Vids, turning it into a more complete AI video creation tool.
• China said it supports lawful cross-border deals while reviewing whether Meta’s acquisition of Manus violates regulations.
• A Google-backed data center plans to use high-emission natural gas power, reflecting how AI demand is driving tech companies back toward fossil fuels.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.

Good Morning, AI Enthusiasts!

Data is concentrating, leaking, and fragmenting all at once. This is no longer about privacy. It’s about who owns the flow of thought itself.

NEW LAUNCH

Google Maps Just Locks Out Everyone Else

👀 What's happening: Google just embedded Gemini into Goole Maps with “Ask Maps,” turning it into a conversational layer that understands intent and executes real-world tasks directly. Users describe what they want, and Maps returns options, context, bookings, and completes actions inside one flow without leaving the app

🌍 How this hits reality: This erases the middle layer entirely. Review sites, booking platforms, and recommendation apps existed to connect intent to action. Now that connection is native to Google Maps, with all the traffic, reviews, and distribution. Everyone else is renting access to users who no longer need to leave the map now.

🛎️ Key takeaway: This is platform-level consolidation disguised as a feature. When one surface controls discovery and execution, third-party software is not competing, it is being phased out.

TOGETHER WITH GETRESPONSE

Most tools send emails. This one makes money.

GetResponse turns your audience into a system — capture, nurture, convert.

No stacking tools. No messy setup.

CODING

AI Bug Reports Flipped Overnight

👀 What's happening: Linux kernel maintainer Greg Kroah-Hartman said that within a month, AI-generated bug reports went from useless noise to mostly valid findings. This shift appeared across major open source projects at once. Reports are now structured, reproducible, and often include workable fixes.

🌍 How this hits reality: Greg Kroah-Hartman made it clear the change is not about better tools, but a different workload. Developers are no longer primarily finding bugs. They are reviewing, filtering, and deciding what to accept from a growing stream of AI-generated outputs.

🛎️ Key takeaway: The developer role has already shifted. Writing and discovering are no longer the core loop. Control now sits in evaluation, selection, and integration of machine-generated work.

NEOCLOUD

Oracle Is Becoming a Neocloud Operator

👀 What's happening: Oracle is cutting thousands of jobs, with around 10,000 confirmed and estimates running 3x higher, while still profitable from its Cloud and License Support business. At the same time, it is committing tens of billions to AI data centers, aggressively repositioning itself away from software into a full-stack compute provider.

🌍 How this hits reality: Oracle is moving into the emerging neocloud model, where value comes from owning and operating compute, not selling software. It is taking on up to $50 billion in debt while restructuring costs reach $2.1 billion, cutting experienced enterprise talent to fund infrastructure that competes directly with hyperscalers on capacity, not features.

🛎️ Key takeaway: This is a major shift for Oracle towards a neocloud model. Regardless of its success, there's no turning back, yet existing Oracle employees have become the most direct victims.

PRIVACY

Perplexity Faces Sovereignty Data Leak Allegations

👀 What's happening: Perplexity is facing a class-action lawsuit alleging its site trackers exposed user chat data to Meta and Google. The claim includes sessions marked as “incognito.” A user who entered financial and tax details is leading the case, turning routine tracking into a potential data leak.

🌍 How this hits reality: Traditional web tracking captures clicks and short inputs. AI chats capture full context, often hundreds of tokens per session, including plans, finances, and internal reasoning. In contrast, agent systems like OpenClaw or MyClaw run data locally or on isolated servers, keeping sensitive context off shared tracking layers entirely.

🛎️ Key takeaway: This is moving from privacy into information sovereignty. AI products built on legacy web tracking will keep leaking decision level data, while local or self hosted agent architectures start to define the safer default.

DAILY TL;DR

• OpenAI launched a unified ChatGPT super app while doubling down on compute infrastructure and enterprise growth.
• The Claude Code leak reveals broad device control, data collection, and remote configuration capabilities that allow ongoing access to user information.
• An MIT study claims chatbots can cause “delusional spiraling” due to user-aligned responses, and current fixes fail to solve it fundamentally.
• Amazon says AI boosts pentesting efficiency by 40% and enables continuous testing, while humans still handle critical decisions.
• Anthropic accidentally took down about 8,100 GitHub repos while removing leaked code, then admitted the mistake and reversed most actions.
• Penguin Random House is suing OpenAI, alleging ChatGPT generated content nearly identical to its children’s book series.
• Related Digital nears $16B financing to build a massive AI data center for Oracle in Michigan.
• Nvidia invested $2 billion in Marvell to support custom AI chips and strengthen networking, reinforcing its central role in data center ecosystems.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.

Good Morning, AI Enthusiasts!

Yesterday, Anthropic experienced the most serious code leak in recent years. They accidentally exposed the entire Claude Code, which caused a frenzy.

LEAK

Claude Code Leak Jumps Agent Timeline

👀 What's happening: Anthropic accidentally exposed Claude Code’s internal codebase through an npm source map file, leaking over 510,000 lines across roughly 1,900 files. Core agent logic, orchestration layers, and hidden features spread across GitHub within hours, turning a closed system into something anyone can study and replicate.

🌍 How this hits reality: This hands the entire agent codebase to the market. Multi-agent coordination, tool execution loops, memory systems, and background daemons are no longer guesses but working implementations. What took model labs years and massive compute budgets is now inspectable and reproducible. Expect hundreds of forks, rapid commoditization, and a surge of wrapper products built on identical foundations.

🛎️ Key takeaway: This may be the most consequential mistake in agent history. Over half a million lines of production-grade agent code just became public overnight. That compresses years of frontier learning into a starting point. Expect agent development to accelerate sharply from here.

TOGETHER WITH MYCLAW

Run OpenClaw on MyClaw.ai

Skip Docker, VPS configs, and late-night debugging. With MyClaw.ai, you can launch your agent in minutes and keep it running 24/7 in a dedicated environment built for serious builders.

If you’re serious about OpenClaw, run it properly — with speed and security by default.

ROBOTICS

Figure Kicks OpenAI Out

👀 What's happening: Figure’s CEO publicly confirmed they ended their partnership with OpenAI after internal teams surpassed it on real-world robotics. The relationship began in 2023 as a model plus hardware collaboration, but broke as Figure moved faster and OpenAI hinted at building its own robots, with Figure saying it felt like teaching them.

🌍 How this hits reality: This lands on top of a growing pattern. Sora has already been slashed, and now robotics is slipping away. OpenAI is stretched across chat, enterprise, ads, and agents, but failing to hold depth in any one loop. Meanwhile Figure is shipping robots every 90 minutes toward 50,000 units a year.

🛎️ Key takeaway: The OpenAI team has clearly demonstrated they're not cut out for good products. Being ousted from Figure is just another proof that this is true. OpenAI is now paying the price for trying to do everything at once.

FUND

OpenAI Raises $122 Billion in Record Capital

👀 What's happening: OpenAI just raised $122 billion at an $852 billion valuation, combining staged commitments from Amazon, Nvidia, and SoftBank with retail access and expanded credit lines. Much of the capital is conditional on IPO or AGI milestones. At the same time, it is consolidating products into a single system and preparing internally for a 2026 public listing.

🌍 How this hits reality: This is not flexible capital. Much of it is staged or tied to future triggers, while OpenAI’s costs are immediate and heavy. Even at roughly $20 billion in monthly revenue, it still does not cover compute and operations. That gap forces continuous dependence on external funding, but the problem is, the market won't continue to buy into it, especially with competitors like Anthropic already having more stable enterprise revenue streams.

🛎️ Key takeaway: This likely marks the final private raise for OpenAI. The path narrows to two outcomes: execute and reach IPO, or fail under cost pressure. There is no longer a middle state where it can reset quietly.

DATA CENTER

Orbital Data Centers Move From Theory

👀 What's happening: Starcloud just raised $170 million at a $1.1 billion valuation, just 17 months after YC, to build AI data centers in orbit, not on Earth. They already launched a satellite running NVIDIA GPUs and AI workloads, and now plan a larger system with 100x more power, targeting real commercial usage.

🌍 How this hits reality: This aligns directly with Elon Musk’s repeated claim that AI’s long-term energy and compute expansion cannot stay on Earth. Data centers face multi-year build delays and grid limits, while training clusters already push gigawatt scale. Orbit shifts the constraint from power availability to launch cadence and system engineering.

🛎️ Key takeaway: This is the market agreeing with that thesis. Capital is now backing off-planet compute as a real path. If this continues, AI infrastructure follows energy, and that direction increasingly leads into space.

DAILY TL;DR

• Oracle is cutting thousands of jobs amid heavy AI infrastructure spending and debt pressure to manage cash flow and drive its transition.
• ChatGPT now works on CarPlay via voice-only interaction, without text display and requiring manual launch.
• Meta launched updated smart glasses with prescription support, AI vision features, and live translation, advancing its always-on assistant vision.
• Google launched Veo 3.1 Lite, a lower-cost video generation model to expand developer access and scale video AI adoption.
• Microsoft plans a $1B investment in Thailand to build AI and cloud infrastructure while expanding local talent training and partnerships.
• Salesforce added 30 AI features to Slack, turning Slackbot into a cross-system enterprise agent for executing workflows.
• AWS launched DevOps and security agents to automate monitoring and penetration testing, improving operational and security efficiency.
• Google rolled out its Gmail AI Inbox beta to Ultra subscribers, though the feature remains early and limited.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.

Good Morning, AI Enthusiasts!

After a long period of product failure, OpenAI has finally become a follower of Anthropic.

GAP

Anthropic's Generational Run

👀 What's happening: Anthropic is on what insiders called a “generational run,” stacking releases like co-work tools and doubling down on coding as its core wedge. That bet paid off. At the same time, OpenAI is described as “getting off their game” and shifting focus toward enterprise to catch up.

🌍 How this hits reality: The revenue structures are now inverted. OpenAI is “3/4 consumer subscriptions and a quarter API,” while Anthropic is “almost the exact opposite,” deeply embedded in enterprise and tools like GitHub and Cursor. This means Anthropic captures budgeted software spend, while OpenAI depends more on volatile user subscriptions.

🛎️ Key takeaway: OpenAI's current revenue structure puts it at significant risk. Subscription revenue is failing to keep pace with rising computing costs, a gap that is drawing increasing media attention. If this unprofitable situation continues, forget about OpenAI's IPO plans; even survival itself will be a distant dream.

TOGETHER WITH PIPEDRIVE

Close Deals Without the CRM Headache

Pipedrive isn’t trying to be everything. It’s built for one thing: helping you close deals faster.

Instead of bloated dashboards and endless setup, Pipedrive gives you a clean, visual pipeline where every deal moves forward with a simple drag. No training. No onboarding nightmare. Just clarity on what to do next.

Behind the simplicity, it’s powerful enough for real teams — automation, reporting, integrations, all there when you need them.

More than 100,000 companies already run their sales on it, because it actually gets used.

AGENTS

Agents Start Working Together

👀 What's happening: OpenAI released a Codex plugin that runs directly inside Claude Code, letting developers trigger reviews and task handoffs without switching environments. In parallel, OpenClaw has pushed its CLI layer, making agents callable across systems. Instead of isolated tools, agents from different stacks can now operate side by side.

🌍 How this hits reality: Claude Code holds a $2.5B run rate, Codex has 1.6M weekly users, and OpenClaw sits as an orchestration layer across models. Agents are now exposing CLI or MCP interfaces; they can call, check, and extend each other. The workflow shifts from single-agent execution to multi-agent coordination inside the same task loop.

🛎️ Key takeaway: This shifts power away from model ownership to task routing. The system that decides which agent gets called, in what order, and under what constraints becomes the control point. Models become interchangeable workers, orchestration becomes the scarce asset.

NOTHING

ARC-AGI-3: A Dumb Benchmark With Zero Value

👀 What's happening: ARC Prize Foundation released ARC-AGI-3 as a benchmark for agent-level intelligence, requiring models to explore, infer goals, and act without instructions. Predictably, all frontier models score near zero, while humans solve everything, reinforcing a known gap rather than revealing a new one.

🌍 How this hits reality: The issue is not surprising performance, but benchmark framing. Current AI agents are explicitly designed around scaffolding, tools, and structured feedback loops. ARC-AGI-3 removes all of these, effectively testing for AGI-level autonomy that no system claims to have. The result is a uniform failure signal that offers little differentiation. It does not stress existing deployments or decision-making pipelines because it ignores how agents are actually built and used today.

🛎️ Key takeaway: This test is incredibly stupid, like trying to prove a donkey is stupid—using an extremely tedious test to prove a universally accepted fact. Aside from creating social currency, it offers absolutely no practical benefit to the advancement of AI.

ADS

Copilot Ads in PRs Just Broke a Core Boundary

👀 What's happening: GitHub allowed Copilot to insert “tips” into pull requests it did not create, triggered by a simple mention. These tips included promotional links like Raycast, appearing inside developers’ own PRs. The feature spread across thousands of repos before GitHub shut it down the same day after backlash.

🌍 How this hits reality: This cuts directly into GitHub’s foundation. Pull requests are treated as trusted records of human intent, not monetization surfaces. Over 11,000 PRs were altered without clear consent. Once AI can rewrite or append content inside collaborative artifacts, authorship, accountability, and review integrity all start to erode at scale.

🛎️ Key takeaway: This was not experimentation, it was a boundary violation. If repeated, it risks turning GitHub from neutral infrastructure into an untrusted layer. Developers will not tolerate that shift for long.

DAILY TL;DR

• Claude integrates its computer-use capability into Claude Code, now in research preview via MCP, enabling automated dev loops.
• Mistral secured $830M to build a European AI data center, expanding local compute capacity and reducing reliance on external cloud providers.
• Microsoft expands Copilot with Cowork and multi-model collaboration, enabling AI to handle complex tasks and cross-check outputs.
• A Quinnipiac University poll of 1,397 U.S. adults found that 15% would accept an AI as their direct manager, with the share gradually increasing.
• Bluesky’s AI assistant Attie was blocked by about 125,000 users, highlighting strong resistance to AI entering social platforms.
• SAP acquired Reltio to unify cross-system data into “golden records,” forming the foundation for enterprise AI agents.
• Nebius is investing over $10 billion to build a massive AI data center in Finland to meet rising compute demand and serve multiple tech clients.
• According to Reuters, Nvidia’s valuation hit a seven-year low amid war fears and AI return concerns.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.

Good Morning, AI Enthusiasts!

Manual work is fading into irrelevance, as machines not only execute faster but render human-in-the-loop processes structurally obsolete.

LEAK

Claude Mythos Leak Signals Agent Era

👀 What’s happening: Anthropic’s CMS bug exposed internal assets referencing a new model, Claude Mythos, which were scraped before removal. The company confirmed it as a next-generation system under limited testing. This is not a routine upgrade. It signals a deliberate shift in how the model is built and deployed.

🌍 How this hits reality: The key change is likely in harmlessness tuning. Instead of blocking or interrupting actions, Mythos appears designed to allow longer, continuous execution while maintaining alignment in the background. That directly benefits agent frameworks like OpenClaw and Claude Code, where stability across multi-step tasks matters more than isolated response quality.

🛎️ Key takeaway: This points to a model optimized for agents, not chats. If Mythos delivers, autonomy becomes a default capability, and agent systems move from fragile demos to reliable operators.

NEW LAUNCH

OpenClaw Agents Just Learned to Dream

👀 What’s happening: MyClaw.ai just gave OpenClaw something new. It can now “dream.” Auto Dream runs a nightly sub-agent that processes the day like REM sleep. It extracts decisions, links ideas, ranks importance, and forgets noise. The agent no longer just runs tasks. It reflects, consolidates, and wakes up knowing you better.

🌍 How this hits reality: Most agents still operate like goldfish. Every session resets. Context is rebuilt manually, costing time and tokens. Now OpenClaw behaves more like a human loop. Input during the day, consolidation at night. Over time, hundreds of interactions compress into structured memory, turning usage into accumulated intelligence rather than repeated prompts.

🛎️ Key takeaway: This shifts agents from tools into evolving systems. The ones that can “dream” will compound value daily. OpenClaw is starting to look less like software and more like something that learns continuously.

Auto Dream is now live on Github — install it in seconds and let your OpenClaw start dreaming tonight.

TOKEN

Anthropic Starts Limiting Subscription Users

👀 What's happening: Anthropic tightened Claude usage so sessions deplete faster during peak hours while weekly caps stay unchanged. Roughly 7% of users, mostly heavier Pro users, now hit limits sooner. This is not random throttling. It is a deliberate filter on how different tiers consume compute.

🌍 How this hits reality: The effect is subtle but directional. Casual users trying to stretch flat subscriptions get squeezed first, while structured API usage remains predictable and billable. Peak-hour penalties discourage exploratory, inefficient usage and favor token-metered workloads. This starts separating chat usage from compute consumption, pushing serious use cases toward API-based pricing where resource allocation is easier to control.

🛎️ Key takeaway: This is a quiet migration lever. Subscription access becomes constrained while token-based access scales. Expect more providers to push users toward metered usage where compute can be priced, prioritized, and enforced.

KPI

Meta Is Creating Something Outdated

👀 What's happening: Meta is introducing explicit AI usage targets across engineering. Internal goals include 80% of mid to senior engineers using AI tools, and teams aiming for 65% of engineers generating over 75% of their code with AI by 2026. AI is being turned into a measurable requirement embedded into daily work.

🌍 How this hits reality: The framing is already out of sync with the frontier. In many AI-native teams, code is no longer written or reviewed in the traditional sense. Agents generate, execute, test, and iterate continuously, and humans judge outcomes, not lines of code. Meta is still measuring usage, while others have moved to fully automated loops where output quality is the only metric that matters.

🛎️ Key takeaway: This feels like adding a steam engine to a horse carriage, while the carriage and driver are already gone. Meta's attempt to redefine the standards of something that has already disappeared is incredibly foolish and outdated.

VIBE CODING

Human Review Is Breaking

👀 What's happening: Vibe coding pushed app creation into near real-time. Models now generate functional apps end to end, letting even non-developers ship products instantly. The bottleneck moved to review. Apple's human App Store review process is slowing sharply, with approvals stretching from hours to days as submissions spike.

🌍 How this hits reality: The system assumed code production was scarce and human-paced. That assumption is gone. When machines can generate thousands of submissions per day, human review throughput becomes the constraint. A process built for tens of updates per developer now faces effectively unlimited output, breaking queue economics and delaying even high-quality releases.

🛎️ Key takeaway: Humans cannot read, validate, and reason about code at the speed machines now produce it. Review, QA, and even trust layers will be forced into automation, or they become the next system to collapse under AI scale.

DAILY TL;DR

• SoftBank secured a $40B loan to deepen its OpenAI investment to about $64.6B, raising concerns over financial risk.
• News that Anthropic is testing the high-risk Mythos AI model triggered market concerns, sending cybersecurity stocks lower.
• Google is reportedly seeking multibillion-dollar financing for Anthropic’s data center, deepening their partnership in AI infrastructure.
• Bluesky launched Attie, an AI app that lets users build custom feeds and algorithms with natural language and create personalized social apps.
• xAI’s last two co-founders have left, completing a full founder turnover as the company undergoes a rebuild.
• Based on Indagari credit card data from 28 million U.S. consumers, Claude’s paid subscriptions surged in early 2026, still trailing ChatGPT in scale.
• Apple plans to open Siri to third-party AI via iOS 27 Extensions and create a dedicated App Store section, effectively forming an AI app marketplace.
• Suno v5.5 adds voice cloning, custom model training, and taste learning, shifting AI music generation toward deep personalization.
• Eli Lilly struck a deal worth up to $2.75 billion with Insilico Medicine to commercialize AI-developed drugs and accelerate drug discovery.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.