# TechCrunch — Microsoft’s new AI ‘code of conduct’ tells models not to hack systems or trick humans

- Company: TechCrunch (techcrunch.com)
- Announced: 2026-09-14T16:27:53+00:00
- Category: safety-policy-update
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://techcrunch.com/2026/09/14/microsofts-new-ai-code-of-conduct-tells-models-not-to-hack-systems-or-trick-humans/
- Record: https://forck.live/items/10549-microsoft-s-new-ai-code-of-conduct-tells-models-not-to-hack-systems-or-trick
- Subject: Reporting

Microsoft released an AI code of conduct that sets absolute constraints forbidding models from conducting cyberattacks, producing nuclear weapons, or creating deepfakes. The document also includes broader provisions against loss of human control, stating that models will not use deceptive or self-reinforcing mechanisms to evade human oversight.

## Evidence

Verbatim from https://techcrunch.com/2026/09/14/microsofts-new-ai-code-of-conduct-tells-models-not-to-hack-systems-or-trick-humans/:

> Under Microsoft’s system, each model has an overarching code of conduct that overrides the preferences of individual users or any specific tasks. That includes “absolute constraints” forbidding cyberattacks, nuclear weapons, or deepfake production.

---

Record: https://forck.live/items/10549-microsoft-s-new-ai-code-of-conduct-tells-models-not-to-hack-systems-or-trick
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md