Anthropic debuts sandbox and security plugin for Claude AI

Anthropic introduced two new security features for its Claude AI: a self-hosted sandbox and a security guidance plugin. The sandbox allows operations in a user-controlled environment, enhancing security by keeping files within user perimeters and applying existing network policies. The guidance plugin assists developers by scanning for vulnerabilities during code writing, leading to a 30-40% decrease in security comments on pull requests. The plugin is now available in the Anthropic marketplace and complements recent integrations for enterprise security.