Meta's CTO admits the company mishandled the creation of a new 6,500-person AI division in March, and promises better management, clearer communication, and more autonomy for workers to repair morale.

What happened: Meta CTO Andrew Bosworth told employees in an internal memo that the company did an 'atrocious' job rolling out its Applied AI engineering unit, which was formed in March with about 6,500 engineers and product managers. Bosworth said Meta failed to explain the vision, support career growth, or paint a clear picture of how the transition would unfold. In response, the company plans to cap managers at about 20 direct reports each, limit manager changes during restructurings, and provide 'AI coaching' tools to employees.

Why it matters: The memo follows reporting that revealed widespread dissatisfaction within the division, with employees describing menial work and calling conditions 'a gulag.' Bosworth acknowledged that executives lost sight of employees' perspectives while rushing to focus on broader strategy and competing in AI coding tools. The unrest is part of a larger morale problem at Meta following mass layoffs and worker surveillance concerns, with several executives including CEO Mark Zuckerberg posting internal messages acknowledging employees' feelings.

What to watch: Applied AI vice president Maher Saba said employees who were forced to join the team will now be allowed to take other roles within Meta if they can secure them, though the group remains initially focused on increasing coding and agentic capabilities of Meta's frontier AI models and could expand to security, debugging, and product development. Bosworth also noted that compute availability will involve 'tough trade-offs for a while' across different teams.