Research agents need evidence before conclusions
AutoResearch proposes independent evidence review for experiments. The idea is useful; the paper’s title is not a universal guarantee.
arXiv2026-09-17
Hallucination-Free
Latest stories
Our latest coverage, with original sources and clear evidence limits. Use the filters to explore a beat.
One hallucination score cannot tell the whole story
Vectara’s September leaderboard update is a reminder to read the task, scoring method and answer rate before comparing models.
Vectara2026-09-22
A legal AI answer now has to pass the source check
Harvey LAB-AA v1.1 adds hallucination checks to legal-agent scoring. What changed, and what its results can tell you.
Artificial Analysis2026-10-08