A former leader at OpenAI, Jan Leike, has raised concerns about the company’s priorities, stating that safety has “taken a backseat to shiny products.” 

Leike, who resigned last week, shared his views on the social media platform X, emphasising the need for greater focus on the safety and societal impacts of artificial intelligence.

But over the past years, safety culture and processes have taken a backseat to shiny products.

— Jan Leike (@janleike) May 17, 2024

Leike, who led OpenAI’s “Superalignment” team alongside co-founder and former chief scientist, Ilya Sutskever, expressed his disappointment with the company’s direction.

“I joined OpenAI because I believed it was the best place to conduct AI research,” Leike wrote. 

“However, I have been disagreeing with OpenAI leadership about the company’s core priorities for quite some time, until we finally reached a breaking point.”

An AI researcher by training, Leike argued for more emphasis on preparing for future AI models, particularly regarding safety. He described the endeavour of building “smarter-than-human machines” as inherently dangerous and highlighted the significant responsibility OpenAI holds on behalf of humanity. 

“OpenAI must become a safety-first AGI (artificial general intelligence) company,” he asserted..

In response to Leike’s departure, OpenAI CEO Sam Altman expressed his appreciation for Leike’s contributions and acknowledged the need for continued focus on safety.

 “I’m very sad to see him leave,” Altman wrote, “he’s right we have a lot more to do; we are committed to doing it.”

In a joint statement, Altman and OpenAI President Greg Brockman said: “There’s no proven playbook for how to navigate the path to AGI. 

“We think that empirical understanding can help inform the way forward. We believe both in delivering on the tremendous upside and working to mitigate the serious risks; we take our role here very seriously and carefully weigh feedback on our actions.”

Recommended reading

• UK and South Korea Prepare for Next Global AI Safety Summit
• Flirty, Fast, and Multilingual: OpenAI Releases New GPT Model
• OpenAI Contemplates NSFW Content Creation

OpenAI confirmed on Friday that it had disbanded the Superalignment team, integrating its members into broader research efforts. 

This move follows the resignation of Ilya Sutskever, another key figure at the company, who announced his departure last Tuesday after nearly a decade at OpenAI. Sutskever, known for his role in temporarily ousting Altman last fall, cited a new, personally meaningful project as his reason for leaving.

Altman announced that Jakub Pachocki will replace Sutskever as chief scientist, describing Pachocki as “one of the greatest minds of our generation,” and expressing confidence in his ability to advance the company’s mission of ensuring AGI benefits everyone.

Earlier last week, OpenAI also unveiled an update to its AI model, showcasing enhanced capabilities such as mimicking human speech patterns and detecting people’s moods. 

Despite these advancements, Leike’s departure underscores ongoing debates within the AI community about balancing innovation with safety and ethical considerations.