Former AI Industry Insiders Call for 'Right to Warn' on Risks: Urging Enhanced Whistleblower Protections

Former employees of prominent AI companies like OpenAI, Anthropic, and DeepMind, including notable figures like Yoshua Bengio, Geoffrey Hinton, and Stuart Russell, have launched the "Right to Warn AI" petition, urging AI companies to enhance whistleblower protections to allow employees to voice concerns about AI risks. The petition, initiated on June 4, emphasises the importance of enabling employees to raise risk-related concerns both internally and publicly without fear of retaliation or silencing through confidentiality agreements. William Saunders, a former OpenAI employee, highlights the necessity for sharing information about potential risks associated with frontier AI systems with independent experts, governments, and the public. The petition proposes four primary principles, including eliminating non-disparagement clauses, establishing anonymous reporting channels, and protecting whistleblowers from retaliation. This initiative arises amid mounting concerns over AI safety, particularly regarding the deprioritization of safety measures by AI labs amidst advancements towards artificial general intelligence (AGI).