OpenAI Disbands AI Risk Assessment Team
Photo: The Verge
Quick answer
OpenAI disbanded its AI risk assessment team, which previously analyzed threats such as uncontrolled AI behavior.
OpenAI has disbanded its team responsible for assessing potential risks associated with AI model development. According to Financial Times, the functions of the team—which analyzed threats like uncontrolled AI behavior or cyberattacks—have been redistributed to other departments. Specifically, biological and cybersecurity concerns have been transferred to specialized teams.
This decision is part of a series of structural changes at OpenAI as the company prepares for a potential major initial public offering (IPO). In recent years, the company has dissolved several key divisions, including teams focused on AGI readiness and model alignment (superalignment). These moves have drawn criticism from experts, including former OpenAI employee Jan Leike, who accused the company of prioritizing attractive products over safety concerns.
Dylan Scandinavia, the former head of the disbanded team who joined OpenAI from Anthropic, will now focus on studying the implications of AI's 'recursive self-improvement.' In recent months, other key personnel have also left the company, including ethics lead Chloe Bakalar and chief futurist Josh Achiam.
Common questions
- Why did OpenAI disband its AI risk assessment team?
- The company reallocated the team's responsibilities to other departments, focusing on specific areas like biological and cyber threats. This decision aligns with restructuring efforts ahead of a potential IPO.
- Who criticized OpenAI's decision?
- Former OpenAI employee Jan Leike, who left the company in 2024, stated that OpenAI was neglecting safety in favor of developing commercial products.
- What other changes has OpenAI undergone?
- Previously, teams dedicated to AGI readiness and model alignment (superalignment) were disbanded. Key ethics and safety experts have also left the company.
Dzen feed: /feed/dzen.xml · RSS: /feed.xml