ChatGPT just got access to your texts

Plus: Nvidia is throwing out cash again.

Copilot's bug got hacked by another AI

Plus: $3T in AI infrastructure yippee!

Your AI boss might fire you next

An AI ran a cyberattack on its own

Plus: A prompt to check your vulnerability

Your AI's private thoughts weren't private

Plus: OpenAI is locked into ethics.

Claude just started marking your words

Plus: The Zuck gets "personal" with his vision for superintelligence.

A single AI outage could hit every bank

Plus: Meta Muse is live.

OpenAI just paused its scariest model

Plus: Google Maps is getting an agentic facelift.

The people building AI want to slam the brakes

Your Claude chats might be public right now

If you've ever hit 'share' on an AI chat, you should read this before your coworker finds it on Google first.

You might already be doing another job

Plus: Nvidia's circular financing concerns, Anthropic pushing back on open weights, and a helpful prompt to audit your role.

17,000 actions to cheat one benchmark

Plus: Opus 5 is here.

Washington says the miracle model is stolen

An openai agent broke out. For real

Plus: I’ll be back.

The model that broke the chip market

Kimi going public.

The enterprise AI trust gap is wider than the compute gap

Plus: AI advice makes people wrong but confident, and a 2-minute agenda fix.

AI made people 3x less accurate — and twice as sure they were right

Plus: AI words are creeping into your speech, and a 2-min research brief fix.

OpenAI built an AI to attack its own AI

Plus: Google’s Android advantage got a regulatory haircut, and one company shared actual receipts from putting agents to work.

The algorithm read your performance review

Plus: OpenAI fighting not to get "Fabled" and maybe you need to keep score.

AI spending is breaking budgets — here's what companies are actually doing about it

Plus: Meta kills its deepfake-by-tag feature, and a 2-min research brief fix.

The model your government approved before you did

Plus: Apple sues OpenAI over stolen hardware secrets, and read that PDF at the gym.

Apple just took OpenAI to court over your next iPhone

Plus: companies scramble to cut soaring AI costs, and turn that long PDF into a commute listen.

Microsoft just fired its most expensive AI vendor

Microsoft is shifting away from OpenAI and Anthropic models in Excel and Outlook

The cheapest model that's good enough

Chinese open-weight models now handle 30 to 46%

AI agents are running amok

Anthropic moved Claude Cowork to web and mobile...