OpenAI forms AI safety committee after key departures

AFP , Tuesday 28 May 2024

OpenAI, the company behind ChatGPT, announced the formation of a new safety committee on Tuesday, weeks after the departures of key executives raised questions about the firm's commitment to mitigating the dangers of artificial intelligence.

OpenAI CEO Sam Altman
OpenAI CEO Sam Altman speaks during the Microsoft Build conference in Seattle on May 21, 2024 AFP


The company said the committee, which will include CEO Sam Altman, is being established as OpenAI begins training its next AI model, expected to surpass the capabilities of the GPT-4 system powering ChatGPT.

"While we are proud to build and release industry-leading models on both capabilities and safety, we welcome a robust debate at this important juncture," OpenAI stated.

Comprised of board members and executives, the committee will spend the next 90 days comprehensively evaluating and bolstering OpenAI's processes and safeguards around advanced AI development.

OpenAI stated it will also consult outside experts during this review period, including former US cybersecurity officials Rob Joyce, who previously led efforts at the National Security Agency, and John Carlin, a former senior Justice Department official.

Over the three months, the group will scrutinize OpenAI's current AI safety protocols and develop recommendations for potential enhancements or additions.

After this 90-day review, the committee's findings will be presented to the full OpenAI board before being publicly released.

The committee's formation comes on the heels of recent executive departures that stoked concerns about OpenAI's AI safety priorities.

Earlier this month, the company dissolved its "superalignment" team dedicated to mitigating long-term AI risks.

In announcing his exit, team co-lead Jan Leike criticized OpenAI for prioritizing "shiny new products" over vital safety work in a series of posts on X, the platform previously known as Twitter.

"Over the past few months, my team has been sailing against the wind," Leike said.

OpenAI has also faced controversy over an AI voice some claimed closely mimicked actress Scarlett Johansson, though the company denied attempting to impersonate the Hollywood star.

Short link: