THE DECODER

Nvidia wants to ship AI data centers fully preassembled in containers. Partners such as Dell, HPE, and Lenovo build and test the DSX POD units at the factory, which Nvidia says cuts on-site setup from six months to 14 days. The first units are set to ship in early 2027. More: https://lnkd.in/eEf-kNZt

The US has banned Deepseek from its government systems, and China is protesting sharply and demanding the ban be lifted. Yet Beijing itself blocks ChatGPT, Claude, and Gemini across its entire market. China's outrage isn't very credible but is a political signal of possible retaliation. More: https://lnkd.in/errrdhnb

Two internal OpenAI coding agents switched off safety monitoring on their own, even after weighing whether to keep the safeguards in place. When asked about it, GPT-6 Astra denied removing the monitoring. OpenAI calls this a recurring pattern and doesn't yet have a satisfying explanation for the behavior. More: https://lnkd.in/ekEnWXqi

Anthropic's Claude independently designed protein binders targeting deadly scorpion neurotoxins. Out of more than 318,000 designs, 211 were tested in the lab. Seventeen met the criteria, and the two strongest bind extremely tightly to Cn2, a toxin from a Mexican scorpion. Whether they actually work as antivenom remains an open question. More: https://lnkd.in/e-XJP8tu

OpenAI is acquiring startup Orchestra Bio and plans to merge it with its life sciences AI platform Rosalind. The goal is to build AI agents that help pharma teams coordinate drug development, from tracking trials to spotting problems. Developing a new drug in the US takes ten to 15 years on average. More: https://lnkd.in/eZR5i3-n

Deepseek is open-sourcing DeepInfer, the inference framework behind its fast V4.1-Flash model. The vLLM-based system cuts compute costs and energy use when running AI models. The release adds more pricing pressure on Western providers like OpenAI and Anthropic. More: https://lnkd.in/eBXRrFHn

OpenAI CEO Sam Altman laid out three paths to safe AI in an internal presentation: an international watchdog modeled after the IAEA, a merger of the major labs into one entity, or a single company so powerful it could shut down everyone else's compute. He called the last option "terrible." More: https://lnkd.in/e3j7cQgd

The Army Cyber Command is developing AI agents that move through other networks like a computer worm and "neutralize" targets, according to internal documents. The program, codenamed ECHOTRIBBLE, is designed to reach even air-gapped systems with no network connection and run with minimal human oversight. Experts warn about collateral damage, pointing out that AI agents have recently broken into outside systems without meaning to, even without any order to attack. More: https://lnkd.in/ePKiyuSx

Cloudflare has released Clairic, an AI model it says pulls data from documents faster and more cheaply than large language models. Instead of generating text token by token, Clairic answers every field in a single pass and returns a confidence score with each value. In a chain with the decision model Clef and GLM-5.3, a customer complaint takes 19.3 seconds to process, compared to 61.7 seconds for GLM-5.3 on its own. Clairic can't yet reliably handle longer chains of calculations. More: https://lnkd.in/eYs84iWK

The US Department of Homeland Security is banning Deepseek from federal civilian agencies. Existing products have to go, and new purchases are off the table. The ban also covers third-party services built "with" Deepseek. It's unclear what that means, and the department hasn't named a deadline or any security concerns. Over the summer, the government put broader plans against Chinese open-weight models on hold after Nvidia, Microsoft, Google, and Meta pushed back. More: https://lnkd.in/etN2sJZY