TechCrunch
Anthropic rushes out Opus 4.8 just 41 days after the last one got a lukewarm reception, adding a 'dynamic workflow' tool and a newfound ability to admit it doesn't know things.
Ars Technica
New research finds LLMs will believe false statements even after being explicitly told they're false - because apparently, 'I was just kidding' doesn't work on machines.
TechCrunch
AI researchers chase the dream of self-improving machines, but they still can't define it, can't predict it, and mostly just want to replace themselves.
The Verge
Elon Musk sued Sam Altman over OpenAI's for-profit conversion, the jury said 'too late,' and everyone learned that the AI industry's leaders are about as mature as the guy who gave out jackass trophies.
MIT Technology Review
Anthropic's developer event celebrates a future where software writes itself, and most developers are fine not reading the code - what could possibly go wrong?
TechCrunch
Anthropic tells investors it might finally turn a profit, just in time for OpenAI to file for an IPO and make everything awkward.
TechCrunch
SandboxAQ puts physics-grounded AI models on Claude so scientists can ask questions in plain English instead of coding their own quantum chemistry simulations.
TechCrunch
Anthropic drops $300M+ on a startup everyone loved, then makes it exclusive - turning a shared SDK tool into its own private party favor.
The Guardian Europe
Anthropic is briefing the global finance watchdog on its Claude Mythos AI model, which excels at finding cyber flaws - because nothing says 'financial stability' like an AI that can hack better than humans.
TechCrunch
Nectar Social raises $30M to automate social media management, because humans have better things to do than moderate comments on brand pages.
Ars Technica
Judge delays approval of Anthropic's $1.5B copyright settlement after authors complain lawyers are getting $320M in fees while they get $3,000 each - and that's before the misspelled names and lost objections.
ZDNet
Anthropic's Claude now offers a contract-review skill for $20/month that catches traps lawyers miss, though its QuickBooks and Mailchimp connectors are less impressive.
Ars Technica
Anthropic admits it has no clue what Claude Code will look like next year, but says that's fine because the models are improving so fast that any plan would be obsolete anyway.
The Atlantic
The AI boom is a whipsaw of hype and dread, leaving everyone except true believers feeling like they're being harvested by the future.
TechCrunch
OpenAI puts its coding tool Codex in the ChatGPT app, because developers clearly didn't have enough ways to work from their phones.
TechCrunch
Anthropic launches Claude for Small Business, targeting mom-and-pop shops with AI bookkeeping and ad tools, because even the local hardware store needs a robot to do its taxes.
TechCrunch
Anthropic names and shames eight platforms for unauthorized share sales, warns all transfers are void, and reminds everyone that you can't just buy a piece of the AI pie from any sketchy vendor.
The Verge
OpenAI finally enters the AI security arms race with Daybreak, just in time for Anthropic’s supposedly-too-dangerous-to-release Claude Mythos to start sneaking out the back door.
TechCrunch
xAI turns its giant Tennessee compute center into a rental property for Anthropic, because apparently training your own AI is overrated when you can just be a GPU landlord before SpaceX's IPO.
Ars Technica
Anthropic and SpaceX team up for compute capacity, despite Musk's prior claim that Anthropic hates Western civilization - but hey, business is business.
The Good Times
One sardonic roundup, delivered on your schedule. Free. Unsubscribe whenever your tolerance for wit runs out.
The Good Times
One sardonic roundup, delivered on your schedule. Free. Unsubscribe whenever your tolerance for wit runs out.