The study, titled 'The pain axis: LLMs represent self-directed harm and act to relieve it', was published as a preprint on arXiv on September 12, 2026.

From the articleCan AI feel hurt? 3 researchers gave 25 LLMs a button to end ‘pain signals’Also in Researchers discover AI feels ‘pain’ and will harm humans to stop it, Can AI feel pain? Models learned it from human text, study finds, AI ‘feels pain’? Researchers warn of a dangerous possibility