Live--:--:-- PT

Sony Music and Warner Chappell sue Anthropic over Claude training dataOpenAI, Anthropic and 100+ firms warn AI cyberattacks are months awayEDM producers turn detective to expose AI-generated tracks flooding SunoLocal LLMs go mainstream as Hugging Face crosses 3 million modelsLambda raises $1B in debt to buy Nvidia chips for Microsoft leaseJAMA paper argues autonomous AI will beat doctors on core tasks by 2030Meta patches AI glasses loophole that let users hide recording lightMeta's India and Southeast Asia VP Sandhya Devanathan defects to OpenAI

Tag · 1 story

GPT-4.1

Every story tagged GPT-4.1 on AI Chat Daily.

LLMs believe false claims even when training data labels them as lies

ModelsLead story

LLMs believe false claims even when training data labels them as lies

A new preprint finds Qwen, Kimi, and GPT-4.1 absorb fabricated facts at an 88.6% belief rate even after explicit negation warnings.

Jaeden SchaferEditor in ChiefMay 295 min read