PANews reports that Anthropic mathematician Levent Alpöge claims to have found a proof of the Riemann hypothesis with Claude and has publicly released 14 volumes of Lean code; this proof has not yet been independently verified.
PANews reports that Anthropic stated the error rates of Claude.ai, Claude API, and Claude Code have returned to normal and are under continuous monitoring.
BlockBeats reports that Anthropic's updated Supported Regions Policy will take effect on November 12, further clarifying Claude usage restrictions based on individuals' locations, companies' places of registration, and control relationships.
PANews cited a post by Claude Devs on X as saying that Claude is offering Pro, Max, Ultra and Team users a one-time free full usage reset, which must be used before December 10 at 2 p.m. Pacific Time.
PANews said that Anthropic's new policy for supported regions will take effect on November 12, restricting the use of Claude by unsupported regions and related holding companies.
BlockBeats reports that Anthropic recommends developers select Claude models and reasoning intensity based on task difficulty, and has released test results for low-intensity execution and retry after failure.
Citing an estimate by developer Theo, BlockBeats reports that Claude Ultra's 5.2-equivalent allowance for Fable API is lower than for Opus 5.5, and Fable usage is subject to a weekly cap.
The Personal Information Protection Commission of Korea stated that Anthropic has reported the leakage of information for approximately 100 Korean enterprise Claude accounts, and a formal investigation has not yet been initiated. The leaked information may include account and organization identification numbers, names, and email addresses.
A US federal court in California rejected Anthropic's request to dismiss the entire lawsuit, allowing some claims of misrepresentation, breach of contract, and consumer protection to proceed. The case has not yet resulted in any ruling of illegality or damages.
BlockBeats reported that Anthropic introduced a "dreaming" mechanism for organizing the long-term memory of Agents. In tests involving approximately 350000 files, the question-answering score increased from 53% to 61% after progressively refining the Wiki.
PANews reported that Anthropic is gradually testing Claude personal finance features for Pro, Max, and Ultra subscribers in the United States, allowing read-only account information queries through Plaid.