AI insiders raise concerns over corporate practices and technological risks

Last week, David Robinson, OpenAI’s safety leader, resigned, joining a growing list of AI insiders who have left their companies due to ethical concerns. In an essay for The Atlantic, Robinson criticized the industry’s reliance on a 'trial and error' approach. Ed Newton-Rex, who left Stability AI in November 2023, also expressed concerns about the use of copyrighted material in training AI models. He founded Fairly Trained, a nonprofit offering certifications for ethical data practices. In June 2024, 13 current and former employees from major AI firms published a letter warning about the risks of advanced AI, including inequality, misinformation, and potential human extinction. The letter was signed by notable figures like Ramana Kumar and Neel Nanda, along with several OpenAI alumni. The letter highlighted a lack of oversight and called for greater accountability. Signatories included Jacob Hilton from the Alignment Research Center, Daniel Kokotajlo of AI Futures Project, and William Saunders, Carroll Wainwright, and Daniel Ziegler, with the latter now working at Anthropic. Six signatories remained anonymous. These departures and public warnings signal a growing unease within the AI industry about its trajectory and ethical implications.