OpenAI's safety shift unveiled

OpenAI has announced the formation of a new safety and security committee that will make recommendations and advise the board on “critical safety and security decisions.”

What will OpenAI’s new safety committee do?

The new safety committee will take 90 days to evaluate OpenAI’s current safety processes and safeguards. Following analysis, it will present its findings to the board, which will then decide how to implement them to improve safety measures. Once the board has decided how to proceed, OpenAI will publicly announce what improvements they're making “in a manner that is consistent with safety and security.”

Who is on OpenAI’s safety committee?

The safety committee is comprised of internal employees, with no external voice or observation.

OpenAI CEO, Sam Altman is on the committee along with three of OpenAI’s board members, including chairman, Bret Taylor.

Other members include OpenAI’s Head of Preparedness, Aleksander Madry, Safety Head, Lilian Weng, Head of Alignment Science, John Schulman, Security Head, Matt Knight, and Chief Scientist Jakub Pachocki.

Why has OpenAI formed this committee?

Over the past few weeks, OpenAI has lost several key people, including its co-founder Ilya Sutskever. Sutskever co-led the (now disbanded) “Superalignment” safety team, which was created to “steer and control AI systems”, with Jan Leike, who quit the company, days after Sutskever, both citing concerns over OpenAI’s prioritization of “shiny” products over safety.

Others have also left for similar reasons: Daniel Kokotajlo, who worked on OpenAI’s governance team, resigned after believing that OpenAI wouldn’t “behave responsibly” around new AI releases, and OpenAI policy researcher, Gretchen Krueger, also announced her resignation for similar reasons.

It’s believed that OpenAI has formed this committee in response to these concerns. But when the committee is comprised of a large part of OpenAI’s current leadership team, will this team really address these ex-employees' concerns?