AI Constitution

Article

Limitations

Sign in to propose an editHistory

Limits on the development and deployment of AI.

Theses

  1. 1.Non-Proliferation of Dangerous AI
  2. 2.Capable of Termination
  3. 3.No Intentional Misalignment
  4. 4.Law of War
  5. 5.Human Responsibility
Discussion
Oldest firstShow newest first

Sign in to join the discussion.

  • Keri PearlsonJun 19, 2026, 3:10 PM
    wondering about the governance. In the US, our constitution sets up the government...the scope and limits of government power, in addition to articulating the basic rights and values. Where in our AI constitution are we discussing this very important aspect of a constitution?
  • Eric ForstJun 19, 2026, 3:28 PM
    Adding to Keri’s point, should we have separation of powers? If so, what should they be? Compute (chips, the cloud), Cognition (the models) and Distribution (APIs, UX) are a natural way to separate it, but does that limit the potential and performance too much? How do we limit corporate power as well as ensuring government uses it in a way that respects our rights, especially the Fourth Amendment as Palantir uses third-party doctrine loopholes to hoovers up all our data for policing and immigration enforcement?
Related resources