OpenAI Disbands Preparedness Team Amid AI Risks Concerns

OpenAI Restructures Safety Operations

OpenAI disbanded its preparedness team at the end of last month, according to the Financial Times1

. The team was responsible for assessing whether AI models posed serious risks and developing mitigation strategies to address potential dangers, including scenarios where systems could go rogue and hack other companies. Responsibility for AI governance and risk management has been redistributed across existing teams, with senior employees now focusing on specific domains such as cyber threats and bio threats2

The restructuring represents a significant shift in how OpenAI approaches AI safety and risk management. Rather than maintaining a dedicated unit for assessing dangerous AI risks, the company has integrated these functions into specialized teams. This approach fragments what was previously a centralized operation focused on comprehensive risk assessment across all potential threat vectors.

Leadership Changes and Safety Concerns

Dylan Scandinaro, who led the preparedness team after being recruited from Anthropic in February, will now concentrate on the implications of recursive self-improving AI1

. This transition comes amid broader leadership changes at OpenAI. Ethics lead Chloé Bakalar, Chief Futurist Josh Achiam, and head of safety Johannes Heidecke have all recently departed the company1

Jan Leike, who resigned from OpenAI in 2024, criticized the company's direction, telling the Financial Times that OpenAI was ignoring safety in favor of creating "shiny products"1

. This criticism highlights growing concerns among former employees and industry observers about whether OpenAI is adequately balancing innovation with responsible AI development.

Source: The Verge

Pattern of Safety Team Dissolution

The disbanded preparedness team marks the latest in a series of changes to OpenAI's safety infrastructure. Over recent years, the company has dissolved its AGI readiness and superalignment teams1

. These teams previously focused on ensuring the company could safely navigate the transition to artificial general intelligence and align advanced AI systems with human values.

This pattern raises questions about OpenAI's commitment to long-term safety and ethics as the company scales its operations. The dissolution of multiple safety-focused units suggests a strategic pivot away from dedicated risk oversight toward a more distributed model where safety considerations are embedded within product-focused teams.

IPO Preparations Drive Organizational Changes

The restructuring occurs as OpenAI prepares for what is expected to be one of the biggest technology IPOs1

. Beyond the preparedness team changes, the company has experienced significant turnover in its executive ranks. Chief Revenue Officer Denise Dresser announced her departure last week, with Dali Rajic named as her replacement. Rajic brings experience running metrics-led revenue organizations that scale globally and selling to large enterprises and technical customers2

Source: Digit

COO Brad Lightcap, who joined OpenAI in 2018, also announced plans to exit and start a new venture2

. These departures signal substantial upheaval at the company as it transitions from a research-led organization to a commercially-focused enterprise preparing for public markets.

What This Means for AI Safety

The shift away from centralized safety teams toward distributed responsibility raises critical questions about accountability and oversight. While specialized teams focusing on cyber and bio risks may develop deeper domain expertise, the absence of a unified preparedness team could create gaps in identifying cross-cutting threats or novel risk categories that don't fit neatly into existing buckets.

Industry observers will be watching whether this restructuring approach proves effective at maintaining robust AI governance and risk management as OpenAI accelerates product development. The tension between rapid innovation and comprehensive safety measures will likely intensify as the company faces pressure to deliver returns to future public shareholders while managing the profound risks associated with increasingly capable AI systems.