Meta Halted an AI Layoff Wave After Agents Fell Short of the Planned Team Cuts

Meta’s reported plan to remake work around AI agents ran into the gap between an automation target and the technology’s output. Internal documents described team reductions of up to 60%, but the agents did not deliver the expected productivity gains, and Mark Zuckerberg halted a planned second layoff wave on May 19.

The plan, codenamed Project OT, reportedly called for many teams to shrink by as much as 60%. Their remaining human employees would work in small, “talent-dense” groups overseeing virtual AI workers. That is a more specific model than simply giving existing staff assistants: it put AI output at the center of the proposed staffing redesign.

Meta has challenged the broadest interpretation of the reported scale. It said it never planned to lay off 60% of its workforce. The distinction matters: the internal-document account describes reductions affecting many teams, while Meta denies a company-wide workforce-cut target of that size.

The workforce model lost internal support

The proposed shift became a workplace flashpoint when employees came to believe that software tracking mouse clicks and keystrokes was training their AI replacements. Employees filled Meta’s internal network with angry posts, while flyers in U.S. offices mocked the company’s AI initiatives and its treatment of workers.

74%Reported internal sentiment

Meta’s internal sentiment reportedly declined from 74% to 55%.

The promised productivity did not arrive

The employee reaction was only one pressure on the initiative. The agents failed to produce the productivity gains Meta had expected, while investors criticized the company’s large AI budget. In July, Zuckerberg acknowledged that the agent technology had not progressed as quickly as he had anticipated.

Zuckerberg halted the second wave on the evening of May 19, hours before the first round of layoffs was scheduled to begin; the second wave had been planned for November. One account also describes layoffs being canceled and parts of the reorganization reversed. The accounts leave unresolved whether the first planned round ultimately occurred.

The reversal does not settle whether AI agents can take on more work at Meta. It does show that a workforce plan built around agent supervision depends on measurable, dependable output arriving before staffing cuts—not after them.