Security experts flag multiple issues in Claude Code, warning, 'As AI integration deepens, security controls must evolve to match the new trust boundaries'

An AI assistant can quickly turn into a malicious insider, so be careful with permissions.