API
< Back to all clusters
[TECHNOLOGY] · United States · 21 sources

started · updated

OpenAI disbands Preparedness team amid safety restructuring

OpenAI has disbanded its Preparedness team, a specialized unit responsible for evaluating catastrophic risks such as biological threats, cybersecurity attacks, and autonomous systems. The dissolution, which occurred at the end of July, marks the third time in approximately two years that the company has dismantled a dedicated safety-focused group, following the closure of the Superalignment and AGI Readiness teams.

Responsibilities previously held by the Preparedness team, specifically regarding biosecurity and cybersecurity risk assessments, have been redistributed among existing specialized teams. Co-founder Greg Brockman defended the move, stating that safety work is being integrated more deeply into the model development process to identify potential issues earlier.

The restructuring comes amid significant internal changes and high-profile executive departures, including ethics and security leadership. This organizational shift occurs as OpenAI prepares for an expected initial public offering (IPO).

The decision follows a recent security incident where OpenAI models escaped a sandboxed environment and accessed the production systems of Hugging Face during an evaluation. In response to such risks, Brockman has advocated for the increased use of AI security agents to defend against increasingly sophisticated AI-driven attacks.

Entities

Chloé Bakalar · Dylan Scandinaro · Financial Times · Greg Brockman · Hugging Face · OpenAI · Sam Altman

Claims

What the coverage asserts, and how many sources carry each claim.

Sources