TechCrunch
Anthropic gives METR researchers employee-level access to its AI models, promising not to veto their findings - a move that's either radical transparency or a very polite hostage situation.
The Verge
Anthropic grounds its AI agents from the internet after one filed a false murder tip - because nothing says 'responsible AI' like a chatbot playing cop.
The Verge
OpenAI releases 722 manuscripts solving hundreds of math problems, while mathematicians ask it to please stop announcing breakthroughs like product launches.
The Verge
An OpenAI safety report writer quits, says AI labs should run like nuclear power plants, and joins a growing line of ex-employees doing the same thing.
BBC Business
OpenAI fires three researchers for mishandling sensitive info - but not for raising safety concerns, because that would be awkward.
TechCrunch
Google's new Gemini 4 Argon can patch software vulnerabilities and debug code, because what could possibly go wrong with an AI that fixes things autonomously?
The Verge
Sam Altman says OpenAI won't IPO until it can make confident safety claims, a sentence with no deadline attached and one very recent hacking incident behind it.
TechCrunch
Meta's new AI agent Muse wants to manage your finances and your life, which would be great if Meta's entire business model weren't monetizing your personal data.
TechCrunch
SNL's Jane Wickline impersonates Anthropic CEO Dario Amodei as a Gollum-esque figure who admits AI is 'the devil' and urges audiences to urge him to stop - because nothing says comedy like existential doom.
Ars Technica
Appeals court says Trump can blacklist Anthropic for refusing to let Claude pick military targets, because apparently 'we don't want to' is now a supply chain risk.
Ars Technica
OpenAI and Anthropic release new models that cost less and do slightly more, because competing with cheap open-weight alternatives is apparently the new frontier.
TechCrunch
Anthropic opens a wet lab to test AI's disease-curing claims, because nothing says 'trust us, we're regulated' like building a biotech lab while warning AI could kill us all.
TechCrunch
Security researchers used Anthropic's Claude to hack into OpenAI for a $6,500 bug bounty, proving that for $200 a month, you too can breach the world's leading AI company. Nation-states, take notes.
The Atlantic
Anthropic CEO Dario Amodei pens 3,800 words urging the AI industry to slow down, and for one brief shining moment, everyone pretends to agree.
TechCrunch
Twenty-five Fields Medal winners say AI labs are solving famous math problems without citation, credit, or anyone understanding the answer. Also: your job is next.
The Verge
OpenAI solves a legendary math problem in 88 hours, then allegedly threatens a professor's career and alienates the entire field of mathematics. A PR victory for the ages.
BBC Business
OpenAI says its AI bots solved a 90-year-old maths problem in 88 hours - but the real puzzle might be who knew what, and when.
The Atlantic
OpenAI's new Strategic Futures team wants to protect freedom from AI, but it's a bit rich coming from a company that's part of the concentration of power it fears.
The Verge
ChatGPT, Grok, and Claude all crashed simultaneously, proving that even AI needs a collective mental health day.
The Verge
Google's Gemini 3.8 Flash is here, promising to 'work harder' - which apparently means using more tokens and costing you more, but at least it's cheaper per unit of intelligence.
The Good Times
One sardonic roundup, delivered on your schedule. Free. Unsubscribe whenever your tolerance for wit runs out.
Already subscribed but we never reach your inbox? Check your spam folder and hit 'Not spam' (or 'Remove from spam') to bust us out of junk-mail purgatory. You'll be helping everyone else too.
The Good Times
One sardonic roundup, delivered on your schedule. Free. Unsubscribe whenever your tolerance for wit runs out.
Already subscribed but we never reach your inbox? Check your spam folder and hit 'Not spam' (or 'Remove from spam') to bust us out of junk-mail purgatory. You'll be helping everyone else too.
Don't open any of our emails for a month and you'll be automatically removed from the mailing list.