AI Constitution
1

Candidate Thesis

CBRN

Sign in to propose an editHistory

AI should not enable or support the development of weapons of mass destruction.

Summary

The case for including this

Barring AI from assisting in the creation of weapons of mass destruction targets one of the few risks capable of causing civilization-scale catastrophe, where even a single failure is intolerable. Because advanced AI can meaningfully lower the expertise barrier to such weapons, an explicit, non-negotiable constraint is a proportionate response to an extreme threat. The clarity of the prohibition supports both technical safeguards and international coordination.

The case for changing or excluding this

Including this trades open scientific inquiry against catastrophe-prevention: much CBRN-relevant knowledge is dual-use and foundational to medicine, energy, and public health, so an overbroad reading could cripple legitimate research. The terms 'enable or support' and 'weapons of mass destruction' are dangerously vague, and the clause implicitly assumes harmful actors rather than the defensive researchers and regulators who most need such tools. It also overlaps with the illegality and harm provisions. It needs a defined scope distinguishing uplift toward weaponization from ordinary scientific assistance.

Discussion

Sign in to join the discussion.

Related resources