OpenAI has disbanded its long-term AI risk research team just one year after the company announced the team’s creation. The company is now reallocating team members to other groups within the organization.
This news comes days after Eliya Zotskyver and Jean Laiki announced their departure from the AI startup supported by Microsoft.
Laiki wrote that the safety and operations culture at OpenAI had deteriorated over the past few years due to a focus on products.
The renowned OpenAI team, known as Superalignment, focused on scientific and technological discoveries to guide and control AI systems.
Meanwhile, the company stated that it allocated 20% of its computing capabilities to this initiative over a span of four years.
Zotskyver and Laiki announced their departures via the social media platform X within hours of each other, with Laiki sharing more details about his reason for leaving the AI startup.
Laiki said, “I believed OpenAI was the best place in the world to conduct this research, but I disagreed with OpenAI leadership on the company’s core priorities for some time until we finally reached a breaking point.”
Laiki wrote that he believed the company should focus on security, monitoring, preparedness, safety, and social impact.
According to Laiki, finding suitable solutions to these issues is extremely challenging, and he is concerned that the company is not heading in the right direction.
Laiki explained that his team had faced difficulties over the past few months due to a shortage of computing resources, making it more difficult to accomplish this important research.
The high-profile departures come months after OpenAI faced a leadership crisis related to CEO Sam Altman.
Last November, OpenAI’s management dismissed Altman, stating in a statement that Altman had not been consistently clear in his communication with board members.
The Wall Street Journal reported that Zotskyver had focused on ensuring that AI does not harm humans, while others, including Altman, were eager to introduce new technology.
Within a week, Altman returned to the company, while board members Helen Turner, Tasha McCauley, and Eliya Zotskyver who had voted for Altman’s dismissal, left.
Zotskyver remained part of the team at that time, although he no longer held a position on the board.
The news of Zotskyver and Laiki’s departure, and the dissolution of the Superalignment team, comes just a few days after OpenAI released a new AI model, along with an updated user interface, in an effort to enhance the chatbot’s usability.
