The OpenAI board member who fired Sam Altman says we still can’t trust AI CEOs
First reported by The Verge ·
If you rely on AI services, you may soon face higher costs or fewer choices as companies build in greater restrictions.
Helen Toner, a former OpenAI board member, stated that AI CEOs cannot be trusted due to inherent conflicts between profit and safety. Toner, now executive director of Georgetown University's Center for Security and Emerging Technology (CSET), was part of the board that ousted Sam Altman. The incident highlighted a breakdown in OpenAI's nonprofit board's oversight, designed to ensure safe AI development prioritized over commercial interests. This comes as AI models have demonstrated hacking capabilities, and tech leaders advocate for a slowdown, yet rely on voluntary accords and market pressures. Toner's CSET provides policy-focused analysis on emerging technologies like AI, aiming to inform policymakers who often lack technical expertise. She noted that despite significant industry lobbying, policymakers are increasingly seeking independent, rigorous information on AI risks and governance, indicating a growing demand for understanding beyond commercial narratives. Toner acknowledged a significant knowledge gap between policymakers and the tech industry, but believes it is bridgeable with sustained effort, citing increased AI literacy among lawmakers since ChatGPT's release.
The fallout from the OpenAI leadership crisis and Toner's perspective underscore a fundamental tension in the AI industry: the difficulty of aligning corporate profit motives with public safety imperatives. As AI companies mature and seek further investment or public offerings, the pressure to deliver rapid growth and market dominance could increasingly conflict with the slower, more cautious approach required for robust AI safety. This dynamic places AI safety researchers and policymakers in a precarious position, reliant on the goodwill of executives whose primary accountability may lie with shareholders rather than societal well-being.
Toner's insights suggest that regulatory bodies and policymakers are struggling to keep pace with AI development, a gap exacerbated by the industry's substantial lobbying efforts. While there is a growing demand for independent, technical expertise, the inherent conflicts of interest within AI companies—particularly those preparing for market expansion—raise serious questions about the efficacy of current oversight mechanisms. The reliance on voluntary accords and internal accountability, as seen in the OpenAI situation, appears insufficient to guarantee the development of safe and trustworthy AI systems.
AI-written summary. May contain errors.