Microsoft publishes an AI 'code of conduct' directing its models to support humans and not deceive or hack systems
2026-09-14
TechCrunch reports Microsoft published a new AI 'code of conduct' laying out general principles its AI models should uphold — including supporting humans rather than replacing them and accelerating human flourishing — alongside specific safety constraints such as not hacking systems or tricking humans.
Significance 3: A major AI vendor issuing a formal, named governance document on model conduct is a credible strategy/policy signal, though it is a principles statement rather than a technical or enforcement mechanism.
Implications · machine-drafted, not owner judgment
A written code of conduct sets a public benchmark other vendors' stated-vs-revealed behavior can be measured against — the real test is what Microsoft does when the code conflicts with a commercial priority, not the document itself. It gives the owner language to point to when advising clients on what governance commitments to demand from AI vendors.
- A documented case where Microsoft enforces (or fails to enforce) this code against its own product roadmap
- Other major labs publishing comparable formal conduct codes in response
Sources
- press Microsoft's new AI 'code of conduct' tells models not to hack systems or trick humans retrieved 2026-09-14