A group of former employees from prominent artificial intelligence (AI) developers are calling for improved whistleblower protection within these pioneering companies. Their goal is to enable employees to voice concerns about the potential risks associated with the advancement of sophisticated AI systems to the public.
On June 4, thirteen former and current employees from OpenAI (ChatGPT), Anthropic (Claude), and DeepMind (Google), along with AI experts Yoshua Bengio, Geoffrey Hinton, and Stuart Russell, initiated the “Right to Warn AI” petition. The petition seeks to establish a commitment from leading AI companies to allow employees to raise risk-related concerns about AI internally and publicly.
One supporter of the movement, William Saunders, a former OpenAI employee, expressed the need for channels to share information about risks associated with new and potentially dangerous technologies to independent experts, governments, and the public.
The petition outlines four main propositions for AI developers. The first proposition is to eliminate non-disparagement clauses that prevent employees from raising concerns about AI risks or subjecting them to punishment for doing so. The second proposition is to establish anonymous reporting channels for individuals to raise concerns about AI risks, fostering an environment conducive to open criticism. The third proposition is to protect whistleblowers, ensuring that companies do not retaliate against employees who disclose information exposing serious AI risks.
According to Saunders, these proposed principles provide a proactive approach to engage with AI companies in order to achieve safe and beneficial AI, which is crucial.
The petition comes in response to growing concerns over the downplaying of safety measures in AI labs, particularly in regard to artificial general intelligence (AGI). AGI aims to develop software that possesses humanlike intelligence and the ability to self-teach.
Daniel Kokotajlo, a former employee of OpenAI, left the company due to a loss of hope in their responsible actions, specifically regarding AGI development.
In a podcast interview on May 28, Helen Toner, a former board member at OpenAI, claimed that CEO Sam Altman was allegedly dismissed from the company for withholding information from the board.
In related news, sci-fi author David Brin suggests using AI to monitor and regulate other AI systems in order to prevent an AI apocalypse.