Останні новини та аналітика про Клод | Yellow

Anthropic said Claude expresses different values depending on the model and language a user picks, after analyzing 309,815 conversations across 20 languages. Key Points: Anthropic sampled 309,815 anonymized Claude.ai conversations covering subjective tasks across three models and the platform's 20 m

Anthropic is on course to post more than $1 billion in profit in Q3 2026, according to an analysis, as the company's confidential IPO filing already in progress. According to SemiAnalysis, Anthropic's annualized recurring revenue is projected to reach $300 billion by end of 2027. At current software

Nous Research released Hermes Mixture of Agents 2.0 on Sunday, which combines outputs from multiple large language models, including GPT, Claude, and DeepSeek, to produce responses that outscore any individual model on standard benchmarks. According to a report, the MoA 2.0 is an update to Nous Rese

Anthropic's Claude Code secretly embedded hidden markers to flag users linked to 147 Chinese domains and AI labs, developers disclosed this week. Key Points Claude Code encoded proxy and timezone details into invisible Unicode markers hidden in system prompts, developers found The mechanism checked

The Trump administration lifted export controls on Anthropic's two most advanced AI models Tuesday, clearing the way for global access to return this week. Key Points The Commerce Department withdrew restrictions on Claude Mythos 5 and Claude Fable 5 roughly two and a half weeks after imposing them.