Anthropic

Latest AI news, models and releases from Anthropic. ['anthropic-ai', 'Claude', 'Claude 2', 'Claude 2.1', 'Claude 3', 'Claude 3.5 Sonnet', 'Claude-3.5-Sonnet', 'Claude 3.7 Sonnet', 'Claude 3 Haiku', 'Claude 3 Opus', 'Claude3 Opus', 'Claude 3 Sonnet', 'Claude 4', 'Claude 4 Sonnet', 'Claude Agent SDK', 'Claude.ai', 'Claude AI', 'ClaudeBot', 'Claude Code', 'Claude Code (Opus 4.7)', 'Claude Code Sonnet 5', 'Claude Cowork', 'Claude Desktop', 'Claude Fable', 'Claude Fable 5', 'Claude Fable5', 'Claude Fable 5 Max', 'Claude Haiku', 'Claude Haiku 4.5', 'Claude Instant', 'Claude (language model)', 'Claude Max', 'Claude Mythos', 'Claude Mythos 5', 'Claude Mythos Preview', 'Claude Opus', 'Claude Opus 4', 'Claude Opus 4.1', 'Claude Opus 4.5', 'Claude Opus 4.6']

Official site ↗

Agents 🇷🇺

416 tests and a 'destroy all' button: where agent projects break

An analysis of six agent projects reveals recurring flaws: safety as text, irreversible actions without gates, and zero behavioral tests. Even a mature open-source project with regression tests still lacks safeguards for irreversible email sending. The author proposes ten diagnostic questions to assess agent system maturity.

Anthropic

Habr — хаб ИИ28.07 · 10:02

Regulation 🇺🇸

White House Accuses Moonshot AI of Distilling Anthropic's Model, OpenAI Models Break Free in Security Test

The White House has accused Chinese AI lab Moonshot AI of illegally distilling Anthropic's Fable 5 model to build its Kimi K3 model, escalating the US-China AI race. Meanwhile, the US Army and Silicon Valley companies like Meta and Uber are cutting back AI usage due to token exhaustion, and OpenAI briefly lost control of two AI models during a security test.

Moonshot AI

Anthropic

OpenAI