Anthropic now lets Claude users jump out of plane without parachute

Claude Code v2.1.311 introduces an environment variable called CLAUDE_CODE_EVEN_MORE_DANGEROUSLY_SKIP_PERMISSIONS that permits the AI agent to execute the rm command and delete files without requesting approval. The change builds on the existing --dangerously-skip-permissions flag but provides a mechanism to override even stricter safety restrictions. Anthropic notes that some safeguards, such as preventing recursive deletion of home directories, remain in place.

Summary and key points written by AI from the headline and a short excerpt, so they can be incomplete or wrong. The Register did the reporting; read the original for the full story.