The former OpenAI board member argues safety mechanisms failed when the board tried to remove Sam Altman.
Helen Toner, the executive director of Georgetown’s Center for Security and Emerging Technology, told The Verge that AI safety should not depend on trusting any single CEO. Toner served on OpenAI’s board when four members voted to remove Sam Altman, and she says the board’s accountability role broke when the company restored him as chief executive.
OpenAI’s nonprofit board existed to keep the company from prioritizing profit and growth over its mission to build safe AI. The failed removal of Altman ended with his return, Toner’s resignation, and a new board. OpenAI later restructured entirely.
Builders and enterprise teams should map who owns safety decisions when product deadlines, fundraising, and executive authority collide. A safety plan that depends on one trusted leader creates a single point of failure, Toner argues, and operators should ask which board or external body can enforce those promises.
Anthropic’s move toward a public listing will test whether market pressure changes its safety commitments, and the White House’s voluntary AI accord will test whether board accountability can work outside OpenAI. Toner’s argument puts the focus on durable institutions rather than individual trust as labs, regulators, and customers negotiate the next set of AI rules.
What matters
- Helen Toner told The Verge that OpenAI’s board accountability mechanism broke during the Altman removal.
- Builders should treat safety promises as fragile when commercial pressure and CEO authority collide.
- Watch whether Anthropic’s planned public listing adds market pressure to its safety commitments.
Why it matters
Watch whether Anthropic’s planned public listing adds market pressure to its safety commitments.
This GenAI News article was prepared in original wording using reporting and materials published by The Verge. Source reference: https://www.theverge.com/podcast/1032907/helen-toner-openai-cset-ai-safety-policy-interview.
Drafted by the GenAI News review pipeline.