← Back to Home

Tag: 安全工程 (1 articles)

How we contain Claude across products

Anthropic detailed their sandboxing techniques for constraining Claude across products, revealing core security engineering practices for building trustworthy AI agents.

Simon Willison · May 31, 2026