Introducing GPT-5.5
OpenAI quietly dropped GPT-5.5 this week, billing it as their “smartest model yet” — faster, more capable, built for complex tasks like coding and research. The naming convention is getting a little fractured (5.5 after 5? Sure, why not), but the capability jump appears real, and dropping this the same week Musk’s lawsuit goes to trial feels like a very deliberate chest-thump.
China’s DeepSeek previews new AI model a year after jolting US rivals
One year after DeepSeek V3 sent American AI stocks into a brief existential tailspin, V4 is here — open source, claiming to rival GPT-5.5 and Gemini, with dramatically lower inference costs and Huawei hardware support baked in. MIT Tech Review has three reasons it matters, but the short version is: China is not slowing down, open source is not losing, and the idea that US export controls have meaningfully kneecapped Chinese AI development is looking increasingly optimistic.
Musk vs. Altman is here, and it’s going to get messy
The trial starts Monday in Oakland, and whatever you think of either man, this is genuinely consequential — it’s fundamentally a question about whether a nonprofit AI safety organization can quietly become one of the most valuable companies in history without anyone being held accountable. The legal merits may be shaky, but the spectacle will be extraordinary, and the discovery documents alone could be worth the price of admission.
How Project Maven taught the military to love AI
In the first 24 hours of the US assault on Iran, AI-assisted targeting enabled strikes on more than 1,000 targets — nearly double the scale of “shock and awe” in Iraq. Project Maven, once controversial enough to trigger employee walkouts at Google, is now so embedded in military operations that it’s reshaping what a modern air campaign looks like. Whatever your politics, the gap between “AI helps you write emails faster” and “AI doubles your bombing capacity overnight” is a chasm worth staring into for a moment.
AI-Designed Drugs by a DeepMind Spinoff Are Headed to Human Trials
Isomorphic Labs — the DeepMind spinoff built on the bones of AlphaFold — says its AI-designed drug pipeline is heading to human trials. This is the part of AI progress I find genuinely thrilling rather than anxiety-inducing: the possibility that a technology that can model protein folding can shave years and billions off drug development. We’ll see if the clinical data holds up, but “AI-designed molecules in human bodies” is a sentence that would have seemed like science fiction five years ago.
The AI Compute Crunch Is Here (and It’s Affecting the Entire Economy)
VC subsidies for cheap AI have masked the real cost of the compute buildout, and 404 Media makes a compelling case that the bill is coming due — showing up in the labor market, gadget prices, and electricity bills. Relatedly, a New Mexico community just voted to deny water to a nuclear weapons AI data center, which tells you everything about how the infrastructure demands of this moment are landing at the local level.
Anthropic created a test marketplace for agent-on-agent commerce
Anthropic ran an experiment where AI agents played both buyer and seller in a classified marketplace, making real deals for real goods with real money. This is either a fascinating research preview of how agentic economies might self-organize, or the setup for a Black Mirror episode — possibly both. Also this week, Anthropic launched Cowork, a Claude Desktop agent for non-technical users built almost entirely by Claude Code itself. The bootstrapping is real.
Why Cohere is merging with Aleph Alpha
Canada’s Cohere is absorbing Germany’s Aleph Alpha with backing from the Schwarz Group (Lidl’s parent company, which is a sentence I did not expect to type today), with explicit government support from both countries. The goal: a sovereign enterprise AI alternative to OpenAI, Google, and Anthropic that isn’t subject to US policy whims. The consolidation of the “not American AI” market into a transatlantic alliance is a genuinely interesting geopolitical move, and probably the right one.
Bottom Line
The week’s news, taken together, is a reminder that AI is no longer a technology sector story — it’s a warfare story, an economics story, a sovereignty story, and a public health story, all at once.