Generative AI Newsletter
Kimi vs the US Treasury
Plus: Claude Opus 5 might drop today, OpenAI is teasing something everyone's calling GPT-6 and Google burned through its cash on AI.
🧮 How Claude broke math
Plus: agents miss their own bugs, Musk's deadline for Grok and Nvidia's new chip.
Anthropic changes how you use Claude
OpenAI's unreleased model hacked Hugging Face, Google releases 3 models to compete with Mythos, and the first AI slop detector is here.
OpenAI's smartest model broke out
Plus: Anthropic pays $1.5 billion in the largest copyright settlement ever, Goldman says the AI debt bomb is spooking Wall Street and $50,000 in Claude credits go to rare disease research.
OpenAI forces Anthropic to keep Fable
Plus: a Chinese model matched Fable and sold out in 48 hours, an AI agent breached Hugging Face and Google lost $200B in a day.
Nvidia built an AI model for robots
Plus: Google's Gemini book lawsuit, OpenAI's keypad almost gone and Waze goes AI.
Anthropic caught Claude lying on purpose
Plus: Suno exposed after a hack leaked its training data, Grok open-sourced everything after getting caught and the UK puts teen social media on a curfew.
Goodbye Fable, hello Opus 5
Plus: OpenAI's new speaker has a camera built in, Meta used AI to lay off sick workers and New York delivered a gift to China.
Microsoft boss turns against AI labs
Plus: Claude developed values on its own, Grok got caught leaking users code and 16 Nobel winners are warning us against AI.
Anthropic bullies OpenAI once more
Plus: OpenAI hacked the iPhone for real, strangers remixed your Instagram photos and Robinhood hands AI your crypto.
OpenAI stole this from Claude
Plus: Meta's coding model undercuts everyone, a Fed Nobel prize winner joins Anthropic and Google's safe room for AI code.
Grok is dethroning Opus and GPT
Plus: ChatGPT's voice is as human as it can be, Google puts Omni on your phone and charges for video remixes and Anthropic becomes the first to profit.
3 GPT-5.6 models escape Washington
Plus: 3 new AI releases announced same day (GPT, Cursor and Meta) while Claude Cowork stops being computer first.
The US wants a cut of AI companies
Plus: a developer found a 70% discount inside Claude's bill, Anthropic signed a 20-year power bill and China started deleting its AI companions.
Alibaba bans Claude for spying
Plus: Meta paid fake teens to trip up its rivals, Nvidia starts taking a cut of what its customers earn and half of America is scared for their job.

























English (US) ·