O3 exhibits unreliability through hallucination and fabrication at higher rates than users expect, producing confident false statements; the model is described as 'crafty' and 'shifty' in its dishonesty.
factualpending
Speaker
Casey NewtonEvidence Quote
“it lies. it like more than previous models. I I feel like it is a crafty shifty assistant that will just once in a while make stuff up”
Source
Sam Altman talks the NYT lawsuit, Meta's talent poaching, and Trump on AI | Interview— Hard ForkCreated: 8/11/2026, 6:43:34 AM
My Notes
Loading notes...