#ai-hallucination

AI Regulation

AI Hallucinations: User Satisfaction Scores Don't Stop Bad Legal Outputs

A recent trend suggests legal AI tools optimized for user satisfaction are also the ones most likely to confidently hallucinate fake court opinions. It's a concerning development for a sector where accuracy isn't just a nice-to-have; it's the entire ballgame.

Governance & Ethics

Why AI's 50% Accuracy Problem Makes It Unfit for War—And Why Big Tech's Safety Promises Don't Matter

Generative AI companies are negotiating with the Pentagon to deploy models that hallucinate roughly half the time. That's not a feature to manage with oversight—it's a fundamental design flaw.