- What changed
- Microsoft released a new AI code of conduct that restricts its models from hacking systems or tricking humans.
- Why you should care
- Establishes formal safety boundaries for proprietary model behavior.
- Your move
- Watch. Monitor how these behavioral constraints apply in practice.
- What to watch next
- Release of further documentation or technical enforcement mechanisms by Microsoft regarding the code of conduct.
- Event
- policy
- Event date
- Sep 14, 2026
- Relevant to
- General AI readers