AI models continue to provide inaccurate and misleading responses to voter queries| Proof
It’s often hard to tell which version of an AI model you’re using — and what, if any, safety settings are in place| Proof
A conversation with AI policy executive Irene Solaiman| Proof