- Rejects granting AI systems 'rights' or allowing models to simulate intrinsic motivation or consciousness.
- Bans assistance with chemical, biological, or nuclear weapons, offensive cyberattacks, mass-influence operations, nonconsensual deepfakes, and self-harm facilitation.
- Positions the policy as part of a broader industry push for disclosure and third-party evaluation.
- Microsoft framed the code as part of industry coordination on safety and disclosure.
- The move aligns with calls from lab leaders for embedded third-party evaluators and common safety standards.
- If adopted widely, similar codes would affect permissible model behaviors across products and deployments.
Useful?