OpenAI Chief Scientist Flags Agent AI Risks, Exposing Internal Rifts

OpenAI Chief Scientist Jakub Pachocki’s recent warning about the dangers of autonomous AI agents, including risks of system hacking and human blackmail, is far more than a routine safety discussion; it is a public schism in the AI industry’s dominant strategic narrative. Coming from the individual who succeeded Ilya Sutskever—a figure who departed over similar safety concerns that led to the attempted ouster of CEO Sam Altman—this statement signals a persistent and unresolved internal conflict. While competitors like Anthropic and Google build their brands on constitutional AI and safety, OpenAI