#### Factored Verification: Detecting and Reducing Hallucinations in Frontier Models Using AI Supervision

We evaluate an automated approach for catching hallucinations in paper abstracts, aiming for consistently more trustworthy results.