Anthropic on March 24 launched three features that expand its AI agents’ autonomy and reach: auto mode for Claude Code adds an AI safety classifier that approves routine developer actions, computer ...