Meta AI Engineers Report Harsh Working Conditions Inside New AI Division

Meta’s workplace culture has faced increasing scrutiny in recent years amid repeated rounds of layoffs and the company’s aggressive push into artificial intelligence. Now, a new report from Wired suggests that frustration within Meta’s recently formed Applied AI division has reached a boiling point.

According to the report, tensions became visible this week when an employee allegedly disrupted a livestreamed internal presentation with an expletive-filled outburst directed at a senior Meta AI executive. The incident reportedly left some participants visibly stunned, with one presenter covering their face during the disruption.

Wired reported that the episode reflects growing dissatisfaction within Meta’s Applied AI unit, a three-month-old organisation comprising roughly 6,500 engineers and product managers tasked with supporting the company’s AI initiatives.

A report published by Business Insider last month detailed how many employees learned they were being reassigned to the group through unexpected emails. One employee later described the process on Reddit as feeling “quite random.” According to an internal announcement reviewed by the publication, Meta created the initiative because its AI models still lacked sufficient real-world knowledge to outperform humans in technical tasks such as coding.

The announcement explained that training advanced AI agents requires examples of how people complete everyday computer-based tasks. To gather that data, employees were reassigned to the programme.

In a leaked recording from an internal meeting, CEO Mark Zuckerberg defended the decision to use company employees rather than outside contractors. He reportedly cited the experience of Meta’s Chief AI Officer, Alexandr Wang, who joined the company after the acquisition of Scale AI, and argued that Meta employees generally possess higher technical capabilities than third-party workers.

Several employees told Wired they felt they had little choice in the matter, describing the reassignment as a “join or leave” situation. Many reportedly refer to themselves as “draftees.” Their primary work involves generating coding exercises and problem-solving tasks intended to train Meta’s AI systems.

Some workers described the assignments in harsh terms, with one employee calling the experience “the gulag” and another saying the work had become “soul-crushing.”

Concerns about morale extend beyond the Applied AI group. Reports indicate that more than 1,600 Meta employees have signed a petition objecting to a programme that tracks clicks and keystrokes for AI training purposes. According to Wired, the atmosphere across the company has become difficult enough that Chief Product Officer Chris Cox addressed employee concerns during a recent company call.

The Applied AI organisation is reportedly led by Maher Saba, a longtime Meta executive who previously served as a vice president within Reality Labs, the division responsible for the company’s metaverse efforts. The team ultimately reports to Meta Chief Technology Officer Andrew Bosworth.

Earlier reports suggested that some managers within the unit were overseeing as many as 50 employees, creating additional organisational challenges.

According to Wired, Zuckerberg acknowledged the situation in an internal memo circulated on Friday. He reportedly admitted that recent organisational changes had caused distress among employees and said the company was working to address mistakes. The memo also reaffirmed Meta’s goal of being “the best place for the most talented people in the world to make an impact.”