OpenAI reportedly disbanded its preparedness team
The Verge Terrence O’Brien ● Covered by 3 sources
OpenAI reportedly shut down its preparedness team last month. That’s the group meant to spot serious model risks and figure out how to reduce them.
Based on reporting by The Verge, Terrence O’Brien — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
OpenAI has reportedly disbanded its preparedness team, according to the Financial Times. The move happened at the end of last month, and it removes the group that was meant to look for serious risks in the company’s models and work out how to reduce them.
That team had a blunt-sounding job: judge whether a model could pose real danger, then help plan the response. Think bio risks, cyber risks, and the sort of nightmare scenario where software starts causing problems outside the lab.
Now that work has been split apart. FT says responsibility for areas like bio and cyber has been pushed into existing teams instead of sitting with one dedicated group. That’s a very different way to handle safety work, and not exactly the kind that screams “we’re slowing down to think.”
The change lands in the middle of a company that has already been through a lot of internal churn as it heads toward what is expected to be a massive IPO. Over the last few years, OpenAI has also been steadily tearing down its more reach-led model, including the dissolution of its AG.
My take — AI-written commentary, not fact-checked reporting
This is how “safety” often gets handled once the money starts staring back: first a team, then a process, then everyone says the work is still happening somewhere else. Splitting risk review into existing teams usually means less clout, not more. The industry loves bold talk about guardrails right up until the guardrails need a budget and a spine.
Read more about this at: The Verge
