OpenAI Safety Expert Returns After Months Away

OpenAI is bringing back Aleksander Madry, its former head of preparedness, months after he left the company that makes ChatGPT to look for new opportunities, because he says AI is advancing so quickly that safety and security have become more urgent priorities.

Madry announced his return in a post on X. He said that what he first thought would be a departure from the company had turned into more of a break from work.

"When I left a few months ago, I felt that the most important thing to focus on was making AI more useful to people," Madry wrote. "But we've since entered a different stage of AI progress. As AI becomes more capable, getting its safety and security challenges right must become a central priority."

Madry will return to OpenAI and work with Jakub Pachocki, the company's chief scientist, to help find ways to handle these challenges.

"As AI becomes more capable, I'm focused on questions about how quickly it is advancing, what that progress means, and how these issues should shape OpenAI's plans and the wider field. I asked Aleksander to come back to OpenAI to work with me on these topics. I trust his judgment and am glad to have him on the team again," Pachocki replied to Madry's post on X.

From AI Safety To Economic Impact

Madry announced that he was leaving OpenAI in May. He described his nearly three years at the company as an "experience of a lifetime."

At the time, he said he wanted to focus on how AI could change the economy, including how businesses understand information, make decisions, and work.

Madry joined OpenAI in 2024 and became its head of preparedness. His job focused on finding and addressing risks from advanced AI models that could cause catastrophic harm. CNBC reported at the time that he was later moved to work on AI reasoning.

Before joining OpenAI, Madry led MIT's Center for Deployable Machine Learning and was a faculty co-leader of the MIT AI Policy Forum.

OpenAI Faces Growing AI And Infrastructure Challenges

Madry's return comes as OpenAI expands the computing systems it needs to support more advanced AI. Major investments and changes in leadership are also reshaping how the company operates.

Last week, OpenAI appointed former xAI infrastructure executive Uday Ruddarraju as chief technology officer for computing, after promoting him in July.

Sachin Katti, who previously led the infrastructure team, is preparing to leave that role but will stay at OpenAI as an adviser.

In July, Katti took charge of GPT-Infra. He led work to use OpenAI's models in semiconductor design and to develop infrastructure. He recently said that customers were already testing the models to design central processing units (CPUs) and edge chips.

The work will now be overseen by Richard Ho, who leads OpenAI's chip development efforts.

OpenAI CEO Sam Altman has predicted that the company will spend trillions of dollars on infrastructure to build more advanced AI models and make them available for wider use.

In recent months, several executives who worked on securing, funding, and building data center capacity have left the company.

Chris Malone, who helped manage OpenAI's data center expansion, left the company in August. Earlier this year, Peter Hoeschele, an important figure in OpenAI's Stargate infrastructure project, joined Meta Platforms with two other colleagues.

Photo: Evolf on Shutterstock.com