AI companies write their own rules, leaving out the public and the military

A new analysis finds that when AI companies write their own ethical rules, they shut out public input. This means critical questions about AI's values and military use are decided by corporations, not democratic bodies.

AI companies are trying to define what 'responsible AI' means. This paper shows that their attempts, even when comprehensive, can actually prevent public discussion about AI's core values. It also means AI models can be used in military operations without the ethical constraints the public expects.