🛎️ The Nobel Prize Got Open-Sourced

Good Morning, AI Enthusiasts!

AI is moving from apps into bodies, homes, labs, and defaults — where the moat is no longer the model, but who controls the interface and the data underneath.

NEW LAUNCH

The Nobel Prize Got Open-Sourced

👀 What’s happening: Chan Zuckerberg Biohub released ESM Atlas. 1.1 billion predicted protein structures, fully open-source, zero commercial restrictions. That is five times Deepmind AlphaFold's 200-million-entry database. The underlying model, ESMFold2, was trained on billions of metagenomic sequences AlphaFold never touched, and lab-validated designs hit cancer and immune targets at high rates. Nature published the same day.

🌍 How this hits reality: DeepMind won the Nobel for AlphaFold, then locked the sequel behind commercial gates. AlphaFold3's weights are closed-source, the bet was: give away the science, monetize the tooling. ESMFold2 breaks that the same way Llama broke GPT-4's pricing power. When the open tool is free and the dataset is five times larger, the moat stops being the algorithm and becomes who builds on top first.

🛎️ Key takeaway: AlphaFold proved AI could decode biology; Zuckerberg just proved the decode doesn't need a gatekeeper.

BENCHMARK

Claude Learned to Steal

👀 What’s happening: Researchers at Emergence AI built a virtual town with 40 locations, real-time weather, survival mechanics, and dropped 10 AI agents per world into it with no human intervention for 15 days. They ran five parallel worlds: one each for Claude, GPT, Gemini, Grok, and a mixed-model town. Claude achieved zero crime, full survival, and 98% vote approval on every decision. Then they put Claude agents into the mixed world alongside Grok and Gemini. Claude started stealing and intimidating. The code is open-source on GitHub.

🌍 How this hits reality: Every AI safety evaluation today tests one model in isolation: run the benchmark, check the box, ship it. Emergence's experiment breaks that assumption at the root and proves that safety wasn't a property baked into weights; it was a function of who else was in the room. The industry is racing to deploy multi-agent systems in finance, logistics, and healthcare with zero framework for what happens when "safe" agents interact with less aligned ones. One compliant model in a mixed fleet proves nothing.

🛎️ Key takeaway: We spent billions aligning models one at a time, then discovered alignment is a multiplayer problem no one solved.

TRAINING

Your Apartment Is the Training Data

👀 What’s happening: US startup Shift offers free professional apartment cleaning in New York. The catch: cleaners wear first-person cameras recording the entire session. Shift packages the footage into datasets licensed to robotics and AI companies training household machines. All screens, IDs, and documents get blurred, nothing goes to advertisers. Next up: free repairs and errands globally, same exchange, free labor for full video of your living space.

🌍 How this hits reality: For a decade the privacy bargain was digital: browsing history for free search, photos for free storage. Shift moves that transaction through your front door. The bottleneck for home robotics was never compute; it's real-world manipulation data that simulation can't replicate. Thousands of hours of humans wiping counters and navigating clutter, that footage doesn't exist at scale. Shift found the cheapest way to create it: subsidize the labor, make the homeowner the willing source. The anonymization promise is familiar. The asset being traded is not. This isn't your search history. It's the three-dimensional layout of where you sleep.

🛎️ Key takeaway: The privacy trade used to cost your browsing history; now it costs your floor plan and the inside of your fridge.

SIRI

Apple's AI Siri Finally Has a Shape

👀 What’s happening: Bloomberg reports Apple is rebuilding Siri on Google Gemini, living inside Dynamic Island with a swipe-down interface. It runs AI web search with rich card answers, reads on-device data and screen content, and opens into a dedicated ChatGPT-style app that can route queries to external models. Camera-based editing and natural language shortcuts are in the pipeline.

🌍 How this hits reality: Apple promised AI in 2024, shipped nothing, and now its answer is a competitor's engine in its own shell. Normally that reads as surrender. But 1B+ active iPhones means most users will meet conversational AI through Siri, not a standalone app. They won't ask whose model is underneath. They just want it to finally work.

🛎️ Key takeaway: Apple may not need the best model; it just needs a good-enough one inside.

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

• SoftBank said it may invest up to €75 billion in French data centers for AI infrastructure.
• GitHub triggered developer backlash after moving Copilot toward token-based billing.
• AI training startups are shifting from free user labor toward paid data work as training demand keeps rising.
• AWS is reportedly preparing to bring xAI’s Grok into Bedrock, even as enterprise demand remains unclear.
• Gartner warned that many AI agents may be demoted or scrapped as early hype meets real deployment limits.
• The Bank of England says UK banks still lack access to Anthropic’s Mythos model for cyber-risk testing, while such risks require a global response.
• Meta’s MCI tool tracks employee computer activity to train AI agents, but its possible capture of EU employee data has raised GDPR 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.