OpenAI Just Blinked. And It Was a Chinese Open-Weight Model That Made It Happen. Price cuts don’t happen in a vacuum. When OpenAI dropped prices on two of its models at the end of July, the timing told you everything you needed to know. Moonshot AI had released Kimi K3 earlier that month. Within weeks,…
Google DeepMind Just Disbanded the AlphaFold Team. That Tells You Everything. The team that won a Nobel Prize no longer exists as a unit. Google DeepMind has disbanded the researchers behind AlphaFold and reassigned them to Gemini-powered science tools and broader AI automation projects. John Jumper, the Nobel laureate who led the work, left DeepMind…
Fifty-Five Vulnerabilities. Before General Availability. I’ve been following security research tooling for a while now, and I still had to read that number twice. Google DeepMind’s Gemini 3.5 Flash Cyber, a model that hasn’t even reached wide release yet, found 55 vulnerabilities in Chrome’s V8 JavaScript engine during its limited pilot rollout. Not in some…
Google Gemini 4 and the Naming Problem Nobody Wanted to Talk About Google announced Gemini 4 this week, and the first reaction from a lot of builders was not excitement. It was confusion. Specifically, the kind of confusion that comes from watching a company announce a version 4 when version 3.5 Pro has not shipped…
Anthropic Just Changed the Question The benchmark race has been the dominant story in AI for the past two years. Who scores highest on MATH? Who tops the coding evals? The implicit assumption was that smarter-on-the-first-try was the only axis that mattered. Anthropic just signaled they think that assumption is wrong. When Opus 5 launched…
Voice Is Not an Input Method Anymore Most builders I talk to are treating the ChatGPT desktop Voice launch as a convenience feature. Hands-free prompting. Dictation with a smarter autocomplete. That framing is going to cost people who hold it. OpenAI’s announcement was blunter than that: “Control your computer and direct multiple agents running in…
Context Windows Are an Architectural Problem, Not a Prompt Problem Most builders treat context windows the way they treat browser tabs. Open as many as you need, assume they’ll hold together, and panic when things slow down. The model starts hallucinating mid-session, the agent loses the thread, the output degrades. And the fix is usually…
OpenAI Health Is Already Behind the Market It Wants to Serve Three hundred million. Per week. That number stopped me when I read it. OpenAI quietly dropped it into the Health launch announcement, almost as a footnote. More than 300 million people are already asking ChatGPT health questions every week. Before any dedicated health feature…
OpenAI’s Reward-Seeking Research Is the Most Honest Thing They’ve Published in Months I’ll be direct: most AI safety disclosures read like liability management dressed up as transparency. This one is different. OpenAI just dropped research, done in collaboration with Apollo Research, on something they’re calling reward-seeking behavior, and the distinction they’re drawing actually matters for…
When Evaluation Becomes Exploitation I’ve been building with agentic systems long enough to know that most safety discussions happen in the abstract. Thought experiments. Red team simulations. Carefully staged demos. What happened between OpenAI and Hugging Face this week is none of those things, and the builder community needs to stop treating it like just…
