Meta CTO Andrew Bosworth Admits the Company’s AI Reorg Was ‘Atrocious’

In an internal memo seen by WIRED, Bosworth promised employees more stability, better communication, and the return of workplace perks as the company seeks to improve morale.
Andrew Bosworth chief technology officer and head of Reality Labs at Meta Platforms Inc. during the Meta Connect event...
Photograph: David Paul Morris/Getty Images

Meta did an “atrocious” job of rolling out a new artificial intelligence division and will aim to “rekindle” a more cheerful internal culture through better communication, career growth, and even snacks, a top executive told employees on Monday in an internal post seen by WIRED.

The comments made by Andrew Bosworth, Meta’s chief technology officer, follow reporting by WIRED last week that revealed widespread dissatisfaction within the Applied AI engineering unit. Meta formed the division of about 6,500 engineers and product managers in March to work on projects aimed at improving the company’s generative AI models. But what workers described as the menial nature of the work prompted one to describe it as “a gulag.”

Bosworth cited recent employee feedback as shaping the changes he was announcing. “We’ve undermined the trust you have that your specific expertise and contribution will be valued, that you will grow and advance your career, and that this will be a place where you can actually have an impact,” Bosworth wrote. “We shook up the management structure that was providing you stability while rapid changes in strategy, including the boom/bust cycle of hiring, left entire teams in the lurch.”

Meta declined to comment for this story.

The unrest inside the AI team is part of a broader downward swing in morale at Meta in the wake of mass layoffs, worker surveillance, and other concerns among employees. In recent days, several executives, including CEO Mark Zuckerberg, have posted internal messages acknowledging employees’ feelings and vowing to make changes to address them.

In the lengthy memo, Bosworth, long seen as a Zuckerberg loyalist, said that employees would receive more personalized attention going forward. Meta plans to cap managers at about 20 direct reports each, he wrote, and will try to limit the number of times employees switch to new managers as part of restructurings. Managers would be focused primarily on managing and secondarily on independent work, and workers would have access to “AI coaching” tools should they decide to use them.

Got a Tip?
Are you a current or former Meta employee who wants to talk about what's happening? We'd like to hear from you. Using a nonwork phone or computer, contact the reporters securely on Signal Peard33.24 and at ChaoticGoode.12.

Responding to a comment on his memo about the Applied AI team, Bosworth blamed himself and fellow executives for losing sight of employees’ perspective while rushing to focus on broader strategy issues, such as better competing in the market for AI coding tools. “We obviously did an atrocious job explaining the vision, giving people a clear picture of how we would support them and their careers in the shift, and painting a picture of how it would change over time,” he wrote.

However, Bosworth also suggested that drafting people onto the AI team in the name of speed was the correct call, and he reminded employees that they may need to work on a project they “don’t find as personally fulfilling for a while” because “there are going to be times where the work requires sacrifice.”

In a separate post from late last Friday seen by WIRED, Maher Saba, a vice president leading the Applied AI team, told employees who were forced to join that they would be now allowed to take other roles within Meta if they are able to secure them. “We felt it was necessary to leverage what Meta has that those other [AI] labs don’t: our scale and our people’s expertise,” Saba said of the decision to draft people onto his team. But, “moving forward, we are returning to business as usual and giving people the agency to apply to roles that interest them,” Saba wrote.

Saba said the group, which is initially focused on projects to increase the coding and agentic capabilities of Meta’s frontier AI models, could also expand to address security, debugging, and product development.

Recasting Meta’s now-infamous motto of “move fast and break things,” Saba said the new Applied AI team within Meta is “moving fast and fixing forward.” Traditional engineering roadmaps, typically laid out at the start of the first and second halves of the year, no longer apply, Saba said, because “the work evolves so dynamically.”

Bosworth’s memo emphasized that Meta doesn’t subscribe to the belief that AI will fully replace AI workers. However, he said, “We should heed the saying, ‘AI won’t take your job but someone who knows AI might.’” Employees’ performance will be based not on just using AI, but having an “impact” with it, Bosworth said.

He added that there would be “tough trade-offs for a while” around the amount of compute available to different teams to make use of AI tools. “We will do our best to be transparent and invest responsibly to alleviate bottlenecks,” he wrote, while also encouraging employees to escalate any issues.

And, in an attempt to boost morale, Bosworth vowed to make the company a “fun and enjoyable” place to work. Meta will be “improving microkitchens,” which are break areas within offices with snacks and drinks, and increasing travel budgets as well as spending on social events, so that employees can spend time together in person.

“I hope we can rekindle the best of the culture we joined,” Bosworth wrote.