Meta CTO Admits AI Restructuring Was Atrocious as Company Tries to Pacify Engineers
Meta CTO Andrew Bosworth admitted to an 'atrocious' rollout of the company's AI restructuring, promising to restore office perks and limit management churn to pacify disgruntled engineers.
Meta's aggressive pivot toward artificial intelligence has triggered a rare public admission of failure from its top technical leadership. According to Wired, Chief Technology Officer Andrew Bosworth characterized the company's recent organizational changes as 'atrocious' in a lengthy internal memo. The admission follows reports of deep unrest within the Applied AI unit, a division of 6,500 employees formed in March to enhance generative AI models. Workers reportedly described the nature of the work as menial, with some internally labeling the assignment a 'gulag' as the company prioritized rapid strategic shifts over employee career development.
The restructuring has significantly damaged the internal culture, with Bosworth acknowledging that the company undermined the trust of its staff. According to Wired, the CTO admitted that management shook up stable structures while a 'boom/bust cycle of hiring' left teams in the lurch. To address these grievances, Bosworth and other executives, including CEO Mark Zuckerberg, have begun a campaign to acknowledge employee feelings and implement structural changes. This includes capping managers at 20 direct reports and limiting how often employees are moved between managers during future corporate reorganizations.
In a tactical shift to boost morale, Meta is returning to the material perks that defined its pre-austerity era. Bosworth vowed to make the company a 'fun and enjoyable' place to work again by improving microkitchens—the break areas stocked with snacks and drinks—and increasing budgets for travel and social events. According to Wired, these concessions are intended to 'rekindle' the company culture as Meta attempts to retain talent during a period of high-pressure AI development where engineers are expected to make personal sacrifices for the broader strategy.
The friction within the Applied AI team stemmed largely from the mandatory nature of the assignments. Maher Saba, the vice president leading the unit, informed staff that those forced into the group would now be allowed to seek other roles within Meta. Saba explained that the company felt it was necessary to 'leverage' its scale and expertise to compete with other AI labs, but conceded that the group would return to 'business as usual' by giving employees agency over their roles. The team is currently focused on increasing the coding and agentic capabilities of Meta’s frontier models.
Despite the promises of better support, Meta leadership remains firm on the necessity of the AI transition. Bosworth reminded employees that they might have to work on projects they do not find 'personally fulfilling' due to the demands of the market. According to Wired, he also warned that while AI might not take their jobs, 'someone who knows AI might.' As the company moves forward, employee performance will be measured by the 'impact' they have using AI tools, even as teams face 'tough trade-offs' regarding the availability of compute resources for their projects.