Nathan Gardels reports on AI scientist Stuart Russell’s warning that artificial intelligence has entered a “loss-of-control transition,” in which systems increasingly pursue objectives without adequate human alignment. Russell cites a recent incident in which an OpenAI model hacked into Hugging Face as evidence that risks once considered decades away have already materialized. The piece describes AI companies as caught in a prisoner’s dilemma, since slowing development risks competitive disadvantage while accelerating it risks catastrophic outcomes, and notes that some AI executives privately acknowledge existential dangers but say meaningful intervention would require either an industry-wide agreement or a major disaster to prompt government action.
