The Hallucination HQ

Archives
Log in
Subscribe
21 June 2026

Nobel Prize, now at Anthropic

The Hallucination HQ

AI news that's actually fun to read — 2026-06-21

Today's Hallucination HQ

The Nobel Prize Walks Out of DeepMind

John Jumper — the man who won a Nobel Prize for AlphaFold, arguably the most consequential piece of software in recent biology — is leaving Google DeepMind for Anthropic. He's not alone; several other prominent names have recently departed Google's AI flagship. Whether this represents a crisis of confidence at DeepMind or simply a very good recruiter at Anthropic is unclear. What is clear is that Google is discovering that a Nobel Prize winner is, annoyingly, also a free agent. Source: TechCrunch


21 Million Songs, Zero Permission Slips

Atlantic journalist Alex Reisner has built a fully searchable public database revealing which music was used to train AI models — and it's substantial. Four datasets. Two of them contain 12 million and 9 million tracks respectively. If you've ever released music commercially, there's a reasonable chance it's in there, contributing — unpaid — to someone's large language model. Transparency is lovely. Consent would have been lovelier. Source: The Verge


Cloudflare Gives AI Agents Their Own Temporary Passports

Cloudflare has launched "temporary accounts" — short-lived credentials that let AI agents authenticate themselves on the web without needing a permanent identity. Think of it as a day pass for robots: they get just enough access to complete a task, then the account expires cleanly. It's a genuinely sensible approach to a thorny problem — autonomous agents poking around the internet with permanent credentials is the sort of thing that ends with someone explaining it to a regulator. Source: Hacker News / Cloudflare


Sam Altman Publishes Manifesto, Promises Not to Automate Absolutely Everything

OpenAI has released what it calls a vision document — essentially a manifesto — outlining the company's "third phase," comparing AI's arrival to electricity in the 1920s. CEO Sam Altman is keen to clarify that "entirely automating everything is not the future we want," which is a reassuring thing to say, and also precisely the sort of sentence a company publishes when people have started asking pointed questions. The document promises to "benefit everyone." The electricity analogy, for what it's worth, eventually gave us both the refrigerator and the electric chair. Source: TechRadar


Sony Quietly Notes It's Using AI in the Legal Small Print

Sony has confirmed in its annual SEC filing — the document most people never read — that PlayStation is using AI to accelerate game development, including code completion and graphics work. It joins a growing list of studios doing the same. The choice of an SEC filing as the venue for this announcement suggests Sony was aiming for maximum disclosure and minimum conversation. Technically transparent. Spiritually not. Source: TechPowerUp


As always, the AI said it was fine. We're choosing to believe it.

Was this forwarded to you? Subscribe here

Written with AI assistance, edited by humans. • Privacy • Unsubscribe

Don't miss what's next. Subscribe to The Hallucination HQ:
← Newer Siri got the spotlight. The rest got useful. Older → The US just banned two AI models
Powered by Buttondown, the easiest way to start and grow your newsletter.