OpenAI appears to be going through turmoil. The company has reportedly disbanded its preparedness team, which was supposed to be responsible for assessing the potential risks posed by the company’s AI models. This move allegedly comes amid the AI company goes through a period of crucial changes.
According to the Financial Times, OpenAI dissolved the team at the end of last month. The group was tasked with examining whether advanced AI models may create serious or even catastrophic risks and finding ways to reduce those risks.
AI Safety Responsibilities Moved To Other Teams
After the decision, responsibilities earlier handled by the preparedness team have allegedly been divided among existing teams. Areas such as biological and cybersecurity risks will now be managed by teams working on those issues.
According to the report, Dylan Scandinaro, who joined OpenAI from Anthropic in February to lead the preparedness team, will now reportedly focus on the potential impact of “recursive self-improving” AI.
Notably, the company has previously shut down or changed teams focused on long-term AI risks, including its AGI readiness and superalignment efforts.
Reportedly, several senior employees linked to safety and long-term AI research have also left the company. These include ethics lead Chloé Bakalar, Chief Futurist Josh Achiam and safety chief Johannes Heidecke.
The changes have raised concerns among some former OpenAI employees. Jan Leike, who left the company in 2024, told the Financial Times that OpenAI was putting too much focus on building new products instead of paying enough attention to AI safety. He criticised the company for prioritising “shiny products.”
These changes at OpenAI are not limited to its AI safety teams. The company has also seen a number of senior executives announce their departures in recent weeks.
Denise Holland Dresser, OpenAI’s Chief Revenue Officer, said last week that she would leave the company in the coming weeks to explore new opportunities.
In a LinkedIn post, Dresser said leaving OpenAI was a difficult decision and expressed pride in what the company and her team had accomplished. Brad Lightcap, who had previously stepped away from his day-to-day COO responsibilities, also announced last week that he would leave OpenAI to start a new venture. The latest departures come at a crucial time for OpenAI as the company prepares for what is expected to be a major public offering.
