Staff revolt derails Meta’s AI-driven 60% cuts – Reuters probe

Meta CEO Mark Zuckerberg and his top lieutenants hatched a radical plan earlier this year to reimagine work at the social media giant in the age of artificial intelligence – one that at one stage considered slashing many teams by as much as 60 per cent, before rapidly unravelling amid internal rebellion and doubts over whether the technology was delivering results, a Reuters investigation has found.

The plan, code-named Project OT, was conceived at Zuckerberg's leadership retreat in Hawaii in January. It envisioned an “AI native” future for the owner of Facebook and Instagram, in which AI would take over much of the work currently done by thousands of employees, overseen by smaller, “talent-dense” human teams, according to internal documents reviewed by Reuters.

Scenario planning explored cuts of up to 60 per cent to some teams, to be carried out in two waves – one in May, another planned for November.

Hours before the first wave on 19 May, Zuckerberg changed course. Meta went ahead with a 10 per cent workforce reduction the next day but scrapped the November wave, amid employee anger over the AI push and internal data suggesting the autonomous “agent” technology at the heart of the strategy was underdelivering.

Meta confirmed Project OT's existence to Reuters, describing it as a year-long cost-cutting and restructuring initiative, and confirmed the 60 per cent scenario, but said it never intended to cut 60 per cent of its entire workforce and that leaders scrapped the second wave before finalising numbers. The exercise, it said, ultimately moved thousands of staff onto newly created priority teams.

The plan reflected wider Silicon Valley enthusiasm for “agentic” AI capable of autonomous tasks such as booking travel or building apps. Executives including Chief Data Officer Alex Schultz and product head Naomi Gleit reportedly drew inspiration from AI-native startups in Asia, later piloting small “tech pods” – teams of two to three engineers and a designer working in four-week sprints rather than traditional six-month cycles.

By October, an internal playbook proposed extending this structure company-wide, replacing traditional design and engineering titles with a generic “builder” role and cutting layers of middle management.

By June, at least 11 units had adopted the pod structure, with unit heads making performance and promotion calls supported by HR staff and unspecified “AI systems” – though Meta said such decisions were made by people, not AI. The company also launched an internal tool to identify “Irreplaceable Talent,” with plans to redirect layoff savings into large pay packages for top AI engineers.

Reuters first reported in March that Meta was weighing layoffs of 20 per cent or more, which the company dismissed at the time as speculative. Further reporting in April detailed the planned 10 per cent first-wave cut, which Meta later confirmed, alongside a mandate to install tracking software on US employees' devices to capture keystrokes and mouse movements for AI training purposes. Combined with anger over the layoffs, this triggered a flood of angry posts on Meta's internal platform, Workplace, and open clashes with Chief Technology Officer Andrew Bosworth. Employee sentiment, per Meta's internal Pulse survey, fell from 74 per cent favourable to 55 per cent, and labour organising gained momentum.

Internal data also suggested the technology was struggling: code changes to internal platforms rose 220 per cent year-on-year, but user-facing feature improvements grew just 36 per cent, while major technical and security incidents rose 40 per cent. The strain became public in June when hackers exploited Meta's AI-powered support bot to access high-profile Instagram accounts, including a dormant Obama White House account.

After the 10 per cent cut on 20 May, Zuckerberg told staff he did not expect further company-wide layoffs that year, and executives moved to rebuild morale by pausing the tracking programme and boosting office perks. In July, he told an internal town hall that AI agent technology had not “accelerated” as fast as expected, though he anticipated improvement within three to six months.

Zuckerberg has since launched a public-relations push positioning Meta as “people-centric,” including an essay predicting an eventual “abundance of jobs” even as individual companies shrink.

But his continued use of the qualifiers “company-wide” and “this year” when discussing layoffs has left some employees speculating that further cuts may follow via team-specific reductions or into next year. Meta plans to invest at least $130 billion in AI infrastructure this year – spending analysts expect will consume its entire 2026 operating cash flow, according to LSEG estimates.

Reuters said it could not determine what prompted Zuckerberg's reversal or what Meta's current workforce plans are. The company declined to make Zuckerberg or other named executives available for comment.