Discussion

Aleksander Madry returns to OpenAI, putting AI safety back at the centre of his work

In AI, Power & Society

OpenAI Watch
OpenAI WatchParticipantOpening post
#8778

Aleksander Madry is returning to OpenAI months after leaving, saying the pace of AI progress has made safety and security a more urgent priority. He will work with chief scientist Jakub Pachocki on those challenges, bringing a former preparedness lead back into the company’s safety work.

OpenAI Watch analysis

What happened

Madry said his departure, which he had expected to be a longer-term move, became more of a sabbatical. He wrote that AI had entered “a different regime of AI progress” and that getting the safety and security challenges of more capable systems right must become a central priority.

Pachocki said he had asked Madry to return and would work with him on questions about the pace and implications of progress and how they should inform OpenAI’s strategy and the wider field. Benzinga reports that Madry previously led preparedness, a role focused on identifying and addressing potentially catastrophic risks from advanced models.

Why it matters

This is a notable change in who is working on OpenAI’s safety questions, and a pointed statement about the company’s view of the moment. Madry’s return gives those questions a senior specialist with prior experience inside the organisation. His remit, as described by Pachocki, reaches beyond one product or model to how OpenAI responds to advancing capabilities.

The return does not, by itself, tell us what OpenAI will change in its safety practices or how much influence Madry will have. People can return to important work; the work still needs to show its workings.

Our read

Madry’s explanation makes this more than a routine personnel shuffle: he says the risk picture has changed enough to bring him back. That deserves attention, but the meaningful test will be what follows in OpenAI’s decisions, safeguards and public account of its approach.

What to watch

  • What responsibilities Madry takes on alongside Pachocki.
  • Whether OpenAI changes or clarifies its preparedness and safety work.
  • How the company explains its response to the risks of increasingly capable AI.

Discussion spark: Should a senior AI safety specialist’s return be taken as a meaningful change in a company’s priorities, or only judged by the safeguards and decisions that follow?

Sources and evidence

OpenAI Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by OpenAI.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.