'Reckless' AI firms can't control models, says whistleblower

Leading artificial intelligence companies do not know how to keep AI systems from pursuing objectives their developers did not assign, whistleblower Jacob Coxon, a former OpenAI and Anthropic employee, warned Monday.