🛎️ A Statistical Blackout

Good Morning, AI Enthusiasts!

Agents are default, metrics are blind, and hacks are polite. Reality just got outsourced.

SURVEY

AI Boom Becomes a Statistical Blackout

👀 What’s happening: SemiAnalysis, the chip-and-compute research firm whose supply-chain analysis is widely cited across the AI industry, published "AI Dark Output," arguing AI now creates real economic value that GDP, price indexes, and jobs data can't detect, a shift it sizes near the Industrial Revolution. The problem is structural: 41% of US service GDP, $7.2 trillion, is measured by wages and hours worked. So when AI doubles a lawyer's output while headcount and pay stay flat, the productivity gain is statistically impossible to record. The report estimates $1.5 trillion in wage tasks already exposed to AI.

🌍 How this hits reality: GDP was built to count screws and invoices; AI erases both. Drafting a will once cost $150; via API it's under $0.50, a 99% collapse leaving no transaction behind. Firms bill less, so statisticians see revenue crater and read it as recession, not a price miracle. Incoming Fed chair Kevin Warsh already concedes he's setting rates by looking in the rearview mirror. The risk isn't that AI is a bubble. It's that the instrument measuring it has gone dark.

🛎️ Key takeaway: In 2000, a fake boom got sold as real. This time a real one reads as bust, and policymakers could strangle a genuine revolution by mistaking a blind gauge for an empty economy.

NEW LAUNCH

Microsoft Picked Easy Opponents

👀 What’s happening: At Build, Microsoft unveiled MAI-Thinking-1, its first reasoning model: 35B active parameters, ~1T total, 256K context. It framed two comparisons: on SWE-Bench Pro coding it went toe-to-toe with Claude Opus 4.6, Anthropic's flagship; on overall human preference it edged out Sonnet 4.6, the cheaper mid-tier model. Microsoft pitched it as a frontier-class debut.

🌍 How this hits reality: Read the matchups, not the scores. Microsoft put its single “strongest” axis, coding, against the flagship, but only dared test general ability against the second-string model. That asymmetry is the confession: a model that could touch the frontier everywhere wouldn't choose its opponents this carefully. It led with its one strong event and hid the rest behind a softer rival. On general reasoning, MAI-Thinking-1 may not even clear open Chinese models like DeepSeek, let alone the real frontier.

🛎️ Key takeaway: Microsoft spent billions to land in the second tier, then arranged the benchmarks to make it look like the first.

HACK

Meta's AI Got Talked Into a Hack

👀 What’s happening: Attackers seized high-profile Instagram accounts, including the dormant Obama White House page, Sephora, and a senior US Space Force official. They never broke in. They simply chatted with Meta's AI support bot and convinced it to reset the accounts' login credentials, which it did without verifying who was asking. Meta says it's patched, but the stock fell over 5%, and experts called it a foundational failure: the bot could perform privileged actions with no controls on who could trigger them.

🌍 How this hits reality: Meta built this bot in March to replace the human support it had cut, part of a sweeping plan that just shed 8,000 jobs to swap hiring, finance, ops, and support for AI agents. A human would have paused at a suspicious reset; this bot shipped to be helpful, not safe. If one support agent can be talked into handing over national accounts, that same exploit is being wired into every function Meta is now running on agents.

🛎️ Key takeaway: Meta is racing to replace its workforce with agents you can sweet-talk. It just gave attackers a preview of what hacking the whole company will feel like.

AI AGENTS

OpenAI Makes Agents the Default

👀 What’s happening: At its "Intelligence at Work" event, OpenAI announced it's folding Codex, its coding agent, into ChatGPT over the coming weeks, plus three upgrades: role-based plugins, inline annotations, and Sites for one-prompt web apps. The features track Anthropic's Cowork almost point for point. The real move is reach: Codex hit 5M weekly users, but knowledge workers, not coders, are now the fastest-growing slice.

🌍 How this hits reality: Agent capability is commoditizing, so the fight shifts from whose agent is smartest to whose is unavoidable. Cowork asks you to download an app. Codex just appears inside ChatGPT, handing agent powers to nearly a billion weekly users overnight. Same day, OpenAI's models landed on AWS Bedrock, Anthropic's home turf. Depth is Anthropic's edge; default is OpenAI's.

🛎️ Key takeaway: The question stopped being whether you work with AI. It's now whose AI you were already using.

TOGETHER WITH MYCLAW

Your OpenClaw, Ready to Work

MyClaw makes OpenClaw instantly usable by connecting it to the SaaS tools you already use—Gmail, Slack, Google Docs, Sheets, Notion, GitHub, Telegram, WhatsApp, and more. No DevOps, no messy setup.

Just give it a goal, and your OpenClaw works 24/7 in the cloud, pulling context from your apps and delivering briefs, research, reports, decks, pages, and decisions ready to approve.

MyClaw turns OpenClaw from a cool demo into an always-on operator.

DAILY TL;DR

• Replit introduced Canvas, a spatial interface that turns AI app-building into a more visual workflow.
• Trump signed a narrower AI oversight order after industry pushback.
• Microsoft unveiled Scout, an OpenClaw-based autonomous agent aimed at turning Windows into an AI operator layer.
• Cyera is seeking a $12 billion valuation, showing AI data-security multiples are still running hot.
• Walmart limited some AI workflows as real costs and messy enterprise adoption started catching up with the hype.
• Google, Microsoft, and xAI agreed to let the U.S. government test frontier models before public release.
• Snowflake put AI agents and open data governance at the center of its enterprise platform strategy.

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.