Published
Anthropic opened its IPO roadshow, offering 825m shares at $70 to $78 each for a Nasdaq listing under the symbol ANT.
Google DeepMind said AI agents formally verified the security properties of the code behind the Signal protocol in four weeks, producing more than six million lines of Lean.
A hotter-than-expected October inflation report pushed the US 10-year Treasury yield up 12 basis points to 4.98% and raised the odds of a Federal Reserve rate rise in December.
Source: Yahoo Finance. Latest data: 10 November 2026.
The yield closed at 4.98% on 10 November, up from 4.86% the day before, after the October consumer price index came in above forecasts. It remains below its six-month peak of 5.31% on 5 October and well above the low of 4.37% on 26 June.
Anthropic began marketing 825m Class A shares at $70 to $78 each and applied to list on the Nasdaq Global Select Market under the symbol ANT.
· Anthropic · Reuters · CNBCGoogle DeepMind said a team of AI agents proved key security properties of the compiled code behind the Signal protocol in four weeks, with human experts writing and reviewing the specifications.
· Google DeepMind · WiredOpenAI said it has deployed a central service across its research infrastructure that can stop a training run, evaluation or data job from receiving further model outputs within 35 seconds of a decision to intervene.
OpenAI · Interesting EngineeringOn the Remote Labor Index from Scale Labs and the Center for AI Safety, Anthropic’s Fable 5.2 delivered client-acceptable work on 41.3% of real freelance projects, about twice the next-best systems.
· The DecoderGemini 4 Carbon, released two weeks ago to a group of cyber defenders, is now on the Gemini API at $2 per million input tokens and $10 per million output tokens, and in a preview multi-agent mode in Antigravity.
· GoogleFor a U-statistic of order \(k\) on a sample of size \(n\) drawn without replacement, Hoeffding, Bennett and Bernstein bounds hold with effective sample size \(n/k\) instead of \(\lfloor n/k \rfloor\).
In test-time reinforcement learning with majority-vote rewards, each question locks into one answer early, and the paper derives the probability of each outcome from the starting logits and a kernel probe.
For a natural class of nonconvex functions, SGD with a decaying learning rate needs exponentially fewer gradient queries to find the global minimum than SGD with any fixed small step size.
A no-go theorem: when prices are diffusive and order flow has long memory, propagator estimation returns a finely tuned decaying impact kernel whether or not true impact decays.
David Gamarnik posted a detailed exposition of OpenAI’s proof of the satisfiability conjecture for random k-SAT, one of the central open problems in discrete probability.
· arXivNvidia closed at $168.89 on Tuesday, a second day of losses, after board members hired the law firm Wachtell Lipton to investigate chief executive Jensen Huang’s relationship with a woman US intelligence links to the Chinese government.
Cubist Systematic Strategies, the quantitative unit of Point72, alleges in a Connecticut lawsuit that its former lead research analyst used Anthropic’s Claude and OpenAI’s Codex to extract proprietary source code before joining a competitor.
· Bloomberg · Complaint (via Bloomberg)US consumer prices rose 0.35% in October and core prices 0.30%, both above forecasts, and markets put the chance of a Federal Reserve rate rise in December at 89%.
· Wolf Street · Yahoo Finance