Latest publications
LLM@AnthropicAI
Anthropic Expanded Project Glasswing: 150 Organizations Gained Access to Claude Mythos
Anthropic expanded access to Claude Mythos Preview for government and corporate clients. Project Glasswing is now used by approximately 150 organizations from 15+ countries.
LLM@AnthropicAI
Anthropic Supports Executive Order to Strengthen US AI Leadership
Anthropic announced support for an Executive Order aimed at strengthening US leadership in artificial intelligence.
LLM@AnthropicAI
Anthropic проанализировала 832 аккаунта: традиционная защита не справляется с AI-атаками
Anthropic исследовала 832 вредоносных аккаунта и выяснила: киберпреступники внедряют AI-атаки, которые традиционная защита не может перехватить. Старые техники безопасности становятся неэффективными.
LLM@AnthropicAI
AI Agent Permissions Should Grow With Their Capabilities — How Anthropic Does It
Anthropic published recommendations for managing AI agent permissions. The company proposes evolving access rights alongside the agent's capabilities and using sandboxing to limit dangerous actions.