Meta created an internal initiative earlier this year, reportedly codenamed Project OT, to explore restructuring the company around artificial intelligence technology. The plan examined scenarios in which certain teams would be reduced by as much as 60 percent, with AI systems performing work previously handled by thousands of employees under the supervision of smaller human teams.
Meta confirmed the planning exercise to Reuters but stated it did not proceed with all proposed scenarios. The company indicated that the effort resulted in moving thousands of employees to newly established priority teams. CEO Mark Zuckerberg directed executives to implement changes to management structures, with the first round of layoffs occurring in May. A second planned round of cuts scheduled for November was canceled.
According to internal documents reviewed by Reuters, Meta intended to use AI agents to automate daily workflows and establish what the company called an “AI-native” operating model. The company planned to redeploy some workers to new positions while laying off others, with one HR executive estimating potential headcount reductions of 25 percent or more. Meta also planned to sell AI agent capabilities to external clients, which began in June.
However, the initiative encountered significant obstacles. Internal reports indicated that AI agents undertook “large-scale, disruptive actions” that humans would not typically execute, resulting in a 40 percent increase in major technical and security incidents. Employee time devoted to resolving problems caused by the AI systems increased by as much as 70 percent. Additionally, productivity gains were inconsistent—while internal software changes increased 220 percent year-over-year, new or upgraded features reaching users only increased 36 percent.
Zuckerberg acknowledged in July that AI development had not accelerated as anticipated over the preceding months. The decision to cancel Project OT reflects broader challenges organizations face in determining appropriate applications for AI technology and assessing whether automation genuinely improves operational efficiency compared to human workers.
Article Attribution | Read More at Article Source
Article summary produced by Claude AI