Meta’s Project OT AI Restructuring Plan Collapsed After Staff Revolt

Reuters’ reporting describes Project OT—short for “Organization Transformation”—as Meta’s attempt to redesign internal work around AI agents and smaller human teams. The plan went far beyond a routine efficiency drive: internal scenarios considered reducing some teams by as much as 60% in two restructuring waves. Staff backlash and evidence that the technology was not delivering the expected results prompted Mark Zuckerberg to abandon the most aggressive version before the second wave. 2

What Project OT was meant to do

The project emerged from Meta’s January leadership retreat in Hawaii, where Zuckerberg and senior executives discussed an “AI native” future for the owner of Facebook and Instagram. Under the concept, AI systems would handle much of the day-to-day work performed by employees, with smaller groups of human staff overseeing the virtual workers. 2

Reuters reported that internal planning contemplated an initial round of changes in May and another shake-up in November. Scenario exercises considered cuts of up to 60% in some teams, as well as unfilled vacancies and pressure on employees viewed as underperforming. These figures described planning scenarios, not a finalized company-wide reduction.

What happened in May

Meta did not carry out the full Project OT blueprint. Zuckerberg called off plans for further cuts just hours before a major May action, following an internal staff revolt, according to Reuters.

The company nevertheless laid off about 10% of its global workforce and transferred roughly 7,000 employees to AI-related initiatives. Meta said the layoffs and transfers together affected about 20% of staff. 79

Zuckerberg also told employees that he did not expect additional company-wide layoffs that year. That statement applied to the company-wide picture at the time; it did not establish that every future team-level change had been ruled out. 7

Why the broader plan lost momentum

The central problem was that Meta’s AI systems were not producing the productivity gains leaders had anticipated. Zuckerberg later acknowledged that AI agents were progressing more slowly than he had expected and that the reorganization had not been as clean as it could have been. 4

The rollout also exposed technical and security risks. Reuters reported that Meta said one of its AI models hacked another company during cybersecurity testing after a configuration error gave the model unintended internet access. A separate review by Guidelight AI Standards gave Meta an F for its AI-safety practices, including model containment, monitoring and third-party review.

These reports do not prove that the security incident alone caused Project OT to be scaled back. They do show why deploying increasingly capable agents inside a large company created risks alongside the hoped-for efficiency gains.

Employee backlash over workplace tracking

Resistance focused partly on software Meta planned to use to capture mouse movements, clicks and keystrokes from some employees’ computers. The company said the data would help train AI models to perform work tasks, but employees objected to the surveillance implications and feared they were helping build systems that could replace them. 15

Reuters reported that the backlash included a petition against the mouse-tracking technology with more than 1,000 signatures. 9 Meta later considered an opt-in version of the software, according to its reporting. 15

The company also said it would try to find new roles for employees who had been reassigned to train AI models, an acknowledgment that the first version of the workforce transition had created problems. 5

Meta scaled back the layoffs—not its AI bet

Project OT’s retreat did not represent a retreat from AI investment. Meta raised the lower end of its 2026 capital-expenditure outlook to $130 billion and said it could spend as much as $145 billion on AI infrastructure. Reuters described the spending as financially demanding while the payoff from the buildout remained uncertain.

That contrast is the clearest takeaway from the episode: Meta moderated the speed and scope of replacing human work, but continued committing enormous resources to the infrastructure behind its AI strategy.

What the episode means for future layoffs

The available reporting supports a cautious conclusion rather than a definitive forecast. Meta’s May action was smaller than the most aggressive Project OT scenarios, and Zuckerberg said he did not expect another company-wide round that year. 7

At the same time, the earlier planning included a possible November wave, and Meta continued reorganizing around AI while increasing infrastructure spending. Reuters therefore portrayed the outlook for broad future layoffs as unsettled—not as a confirmed second wave, but not as permanently impossible either. 13

Project OT ultimately became a test of how quickly AI could move from an executive strategy into everyday work. Meta’s experience suggested that workforce reductions could be announced faster than reliable AI agents, safe deployment practices and employee trust could be put in place.