meta cto admits ai reorg was atrocious as morale crisis grips applied ai team

At a glance:

• Meta CTO Andrew Bosworth admitted the company's Applied AI division rollout was "atrocious" and promised reforms.
• The AI division, formed in March with 6,500 engineers, faced worker backlash describing the environment as a "gulag".
• New policies include capping managers at 20 reports and allowing employees to apply for other roles.

Bosworth's candid admission

Andrew Bosworth, Meta's chief technology officer, delivered a blunt assessment of the company's handling of its artificial intelligence reorganization during an internal address to employees on Monday. The comments come on the heels of extensive reporting by WIRED that revealed significant unrest within the Applied AI engineering unit, which Meta established in March to focus on advancing the company's generative AI models.

The division was created by drafting approximately 6,500 engineers and product managers onto the team, a decision that led to widespread dissatisfaction among workers. Many employees described the nature of their work as menial and the work environment as oppressive, with one worker likening it to "a gulag." Bosworth directly acknowledged these concerns in his memo, taking responsibility for the company's failure to properly communicate the vision and support structure for the new division.

In his message, Bosworth emphasized that employee feedback directly shaped the changes he was announcing. He admitted that Meta had undermined trust by failing to clearly articulate how the reorganization would benefit individual careers and contributions. The rapid changes in strategy, combined with what he described as a "boom/bust cycle of hiring," left entire teams in uncertainty, further eroding confidence in leadership.

Structural and cultural reforms

Moving forward, Bosworth outlined several concrete measures aimed at rebuilding trust and improving the work environment. Perhaps most notably, Meta will cap managers at approximately 20 direct reports each, a significant reduction from previous structures that often saw managers overseeing larger teams. This change is designed to provide more personalized attention to employees and improve communication between management and individual contributors.

The company will also work to limit the number of times employees are forced to switch managers due to restructuring efforts. Managers will be instructed to focus primarily on management duties while taking on secondary independent work, allowing them to better support their teams. Additionally, workers will have access to "AI coaching" tools if they choose to utilize them, though participation will be optional.

Beyond structural changes, Bosworth committed to making Meta a "fun and enjoyable" place to work again. The company plans to improve its microkitchens—office break areas equipped with snacks and drinks—and increase travel budgets as well as spending on social events. These initiatives aim to encourage in-person interaction among employees and help rekindle what Bosworth described as the best aspects of Meta's original culture.

Shifting priorities and future expectations

Maher Saba, a vice president leading the Applied AI team, communicated separate updates to employees, including news that those who were initially drafted onto the AI team would now be permitted to seek other roles within Meta if they can secure them. Saba explained that the decision to assemble the team by leveraging Meta's scale and existing talent was initially necessary, but the company is now returning to more traditional hiring practices.

The Applied AI team is currently focused on projects aimed at increasing the coding and agentic capabilities of Meta's frontier AI models. However, Saba indicated the group could expand to address additional areas including security, debugging, and product development. This evolution reflects the dynamic nature of AI development, where traditional engineering roadmaps established at the beginning of each half-year no longer adequately capture the rapidly changing landscape.

Recasting Meta's famous motto of "move fast and break things," Saba introduced the new mantra "moving fast and fixing forward." This shift acknowledges that the work in AI development is so dynamic that rigid planning structures are less effective than adaptive approaches that can respond to emerging challenges and opportunities.

Addressing the AI workforce reality

Despite acknowledging the challenges of the reorganization, Bosworth defended the decision to rapidly assemble the AI team, suggesting that speed was necessary to compete effectively in the market for AI coding tools. However, he also recognized that there would be difficult periods ahead where employees might need to work on projects that don't align with their personal interests.

Bosworth emphasized that Meta does not believe AI will fully replace AI workers, but he warned that the landscape is evolving rapidly. "We should heed the saying, 'AI won't take your job but someone who knows AI might,'" he told employees. Performance evaluations will consider not just the ability to use AI tools, but demonstrated impact achieved through their application.

The CTO also acknowledged that there will be "tough trade-offs" regarding compute resources available to different teams as they integrate AI tools into their workflows. While Meta will strive to be transparent and responsibly allocate resources to minimize bottlenecks, employees will need to escalate issues when they arise.

Broader context of employee concerns

The unrest within Meta's AI team represents part of a larger pattern of declining morale across the company. Recent months have seen mass layoffs, increased worker surveillance, and other concerns that have contributed to a challenging work environment for many employees.

In recent days, several executives including CEO Mark Zuckerberg have posted internal messages acknowledging employee sentiments and pledging various changes to address their concerns. This broader recognition of workplace issues suggests that the AI team's struggles are not isolated incidents but part of systemic challenges facing the organization as it navigates rapid technological change and workforce adjustments.

The combination of structural reorganization, resource constraints, and cultural shifts has created a complex environment for Meta's workforce. As the company continues to invest heavily in AI development while managing employee expectations and retention, the lessons learned from the Applied AI team's experience may serve as a case study for how large technology companies approach major organizational transformations.

Looking ahead

As Meta continues to navigate these challenges, the success of its cultural reforms will largely depend on consistent execution and genuine engagement with employee concerns. The company's commitment to limiting managerial spans of control and providing more flexibility for role transitions represents a recognition that traditional tech industry practices may not be sufficient for retaining top talent in the competitive AI field.

The emphasis on improving physical workplace amenities like microkitchens and social events reflects an understanding that workplace satisfaction extends beyond compensation and career advancement. Creating spaces and opportunities for informal interaction can help rebuild team cohesion and trust that may have been damaged during the initial rollout of the AI division.

Whether these changes will be sufficient to address the underlying tensions within Meta's AI organization remains to be seen. The company faces the ongoing challenge of balancing aggressive AI development timelines with the need to maintain a motivated and productive workforce. Success in this balance will likely prove crucial not just for Meta's AI ambitions, but for its broader ability to attract and retain talent in an increasingly competitive market for artificial intelligence expertise.