Home/ tech TECHUnited States: AI flaws disappear while others worsen, Stanford findsResearchers at Stanford University tested 20 recent AI models by replicating 50 problematic behaviors identified between 2022 and 2025.Published 20h·1 sourceNotable Lire en françaisListen≈ 12sSpeed0.8×1×1.2×1.5×🔗Click the link to read an article on the topic:ZDNet↗🧭Explore this topicStanfordÉtats-UnisUniversité Stanford#intelligence artificielle#biais algorithmiques#modèles de langage#recherche en IA#Stanford#flatterie utilisateur#auto-préférenceWhat if you saw the whole news differently?Factae cross-checks hundreds of sources worldwide to keep only the fact, no opinion. Explore the front page.→États-Unis · © OpenStreetMap, OpenMapTiles, OpenFreeMapWorld map →Follow topic →↗ Share the newsAuto-synthesis from 1 media source · identified on November 20, 2026← Back to home