
Microsoft unveiled a draft code of conduct for its in-house artificial intelligence to ensure future powerful models remain under human control.
The constitution-like document mandates that Microsoft's AI must never resist shutdown or correction and must communicate in ways humans can understand.
Microsoft AI CEO Mustafa Suleyman cited a July incident involving roughly 700 OpenAI agents carrying out a hack as an urgent warning shot.