Meta's plan to replace up to 60% of some team members with AI collapsed after internal data showed the initiative was causing more problems than it solved, according to a Reuters investigation published Wednesday. The company, which had been preparing the cuts under a project it called OT (Organization Transformation), backed off after executives saw a surge in technical incidents, security issues, and employee resistance.
Internal posts reviewed by Reuters showed that code changes to internal software platforms were up 220% year-over-year, but changes that actually reached Meta users grew only 36%. Meta CTO Andrew Bosworth wrote about the discrepancy in early June. Other internal posts flagged "reliability warning signs" caused by the AI coding surge, and an April post said unchecked AI agents were performing "large-scale, disruptive actions that humans are unlikely to execute."
Major technical and security incidents, including service disruptions and possible data leaks, spiked 40% from the previous year. The time staffers spent firefighting those incidents rose 70%.
Employees pushed back
In April, Meta mandated tracking software on US employees' devices to capture keystrokes and mouse clicks, intending to teach AI agents to replicate human-computer interaction. Employees rebelled, believing they might be training their own replacements.
Meta tried to placate workers with promises of increased travel spending, more social events, and better snacks in office microkitchens. None of it appeared to boost morale. According to Reuters, Zuckerberg has stuck to the words "company-wide" and "this year" in discussing layoffs, leading some employees to speculate he will continue cuts through team-specific or performance-based dismissals.
What IT leaders should take from this
Consultants said the Meta story shows what happens when AI marketing hype goes unchallenged. Sanchit Vir Gogia, chief analyst at Greyhound Research, said, "Meta trusted a forecast of AI capability before it existed in production, a different failure from trusting AI too much."
"Meta booked a forecast as capacity," Gogia said. "Agents will improve, but the error was budgeting that improvement as production capacity before it arrived. Prove the action before widening the authority, and the authority before removing the human control. Only then is removing human capacity a decision, not a bet."
Terra Higginson, principal research director at Info-Tech Research Group, said experienced IT leaders should not be surprised. "Unchecked AI agents are a bad idea. Removing humans is a bad idea. That's not the future anyone wants," she said. "We want work to be reimagined so the human part still matters and technology makes it better."
Higginson warned against treating AI output as a productivity measure. "What we are already seeing, though, is lots of output and action without always getting the outcome we actually want. We should not use AI output as a proxy for productivity. Humans bring judgment and friction before taking actions with significant consequences; agents can remove that friction. We don't want easy outcomes, we want good outcomes."
Justin Greis, CEO of consulting firm Acceligence, said the core issue is the gap between activity and value. "AI can make an organization extraordinarily busy without necessarily making it more productive," he said. "If an AI agent produces ten times as much code, analysis, or work product, that does not mean the enterprise created ten times as much value. It may simply mean the company created ten times as much material that somebody now has to validate, secure, integrate, maintain, or clean up."
Greis said the question executives should ask is not "How much work can AI produce?" but "What measurable business outcome improved because AI produced it?"
Why this matters for management
For managers, the Meta case is a warning about measuring the wrong things. Activity metrics - lines of code, tickets closed, output volume - become even less reliable when AI can manufacture them at machine speed. The discipline of tying AI investments to actual business outcomes, and keeping human oversight in place until those outcomes are proven, is what separates useful AI adoption from expensive disruption. For practical guidance on applying these lessons, see AI for Management and AI for Executives & Strategy.
Your membership also unlocks:
