AI Selloff Wipes Out $2 Trillion as Anthropic Cuts IPO Target to $1.2 Trillion
A post-midterm rout erased more than $2 trillion from tech stocks on 6 November, with Nvidia, AMD and Micron among the hardest hit and the losses spreading to Asia. Anthropic trimmed its planned Nasdaq-debut valuation from a $2 trillion figure to $1.2 trillion to $1.3 trillion. Ramp data showing AI spending nearly flat since August suggests the demand question predates the vote.
Read full story →Royal Navy Flies Strike Drones off Carrier as Geneva Readies Autonomous-Weapons Talks
Britain launched four one-way attack drones from HMS Queen Elizabeth for the first time in October, part of a plan to pair cheap expendable drones with its F-35B fighters. The trial lands as diplomats prepare to meet in Geneva this month to decide whether to negotiate rules for lethal autonomous weapons. The capability is arriving faster than the doctrine meant to bound it.
Read full story →Nous Research Releases Arete-27B, a Fine-Tune That Cuts Qwen's Propaganda Answers to 0.9%
Nous Research post-trained Alibaba's open Qwen3.8-27B to drop propaganda-tagged answers from 22% to 0.9% on its own 803-prompt test, saying it aimed to "reinforce western values on a Chinese-trained model." The results come from Nous's own eval graded by a single model, so they show a shift toward chosen values, not neutral accuracy. The fine-tune ships under Apache 2.0.
Read full story →OpenAI Launches GPT-6 Cyber, Says It Found 81 New Firefox Flaws and 948 in Linux
OpenAI released a security model it says has validated thousands of exploits, prepared patches for 948 Linux kernel issues and found 81 novel Firefox vulnerabilities. Access is gated to vetted "Daybreak Red" customers, and OpenAI pledged $1 billion to critical-infrastructure defenders. Most of the figures are OpenAI's own and await independent confirmation from the affected projects.
Read full story →