AI agents & automation
10 August 2026·5 min read
Claude Code switches to autonomous mode by default: the lesson for how you supervise your own AI agents
From August 14, 2026, Claude Code stops asking for approval before every action on Pro, Max and Team plans — a safety classifier decides instead. Anthropic’s own numbers on "approval fatigue" say something important about how any SME should supervise the agents it deploys.
Read →