Employees at Meta have taken an unusual step. They printed flyers. They placed them in bathrooms, on vending machines, and in meeting rooms across U.S. offices. The message cut straight to the point. “Don’t want to work at the Employee Data Extraction Factory?”
The pamphlets appeared on Tuesday, just days before a major round of layoffs. They urged colleagues to sign an online petition at mcipetition.com. The documents cited protections under the National Labor Relations Act. Workers are legally protected when they choose to organize for the improvement of working conditions. But this wasn’t a standard labor dispute over pay or hours. It targeted software quietly installed on company computers. That software records mouse movements, clicks, keystrokes, and occasional screen snapshots.
Meta introduced the tool in late April. Internal memos described it as the Model Capability Initiative, or MCI. One note, posted by an AI research scientist in a channel for the company’s Meta SuperIntelligence Labs team, declared, “This is where all Meta employees can help our models get better simply by doing their daily work.” The goal sounded straightforward. Capture real human interactions with software interfaces. Feed that data into systems designed to build autonomous AI agents.
Meta spokesperson Andy Stone explained the logic in comments to Reuters. “If we’re building agents to help people complete everyday tasks using computers, our models need real examples of how people actually use them — things like mouse movements, clicking buttons, and navigating dropdown menus.” The data, according to the company, would not support performance evaluations. Safeguards would protect sensitive content. Yet many staffers saw something different. They saw themselves forced to train the very systems positioned to take over their roles.
The timing sharpened the resentment. Meta plans to cut about 10% of its workforce, roughly 8,000 jobs, with the first wave scheduled for May 20. Additional reductions will follow later this year. The company had confirmed the sweeping layoffs more than a month after Reuters first reported them. Executives have tied the reductions to rising costs for artificial intelligence investments. Employees, however, connected the dots. Surveillance today. Replacement tomorrow.
From Quiet Frustration to Open Defiance
Anger had brewed for months on internal platforms. Staffers complained about the push to adopt AI agents for coding and routine tasks, even when those tools slowed them down initially. Meta CTO Andrew Bosworth outlined the long-term vision in a memo. “The vision we are building towards is one where our agents primarily do the work and our role is to direct, review and help them improve.” He spoke of agents that would “automatically see where we felt the need to intervene so they can be better next time.” The company would stay rigorous about collecting data and evaluations for every type of interaction.
But rigor from the top met resistance from below. The flyers marked the most visible sign yet of a budding labor effort inside one of the world’s largest tech firms. In the UK, some Meta employees have already begun organizing with the United Tech and Allied Workers, a branch of the Communication Workers Union. Organizer Eleanor Payne didn’t hold back. “Meta’s workers are paying the price for management’s reckless and expensive bets,” she told TechRadar. “While executives chase speculative AI strategies, staff are facing devastating job cuts, draconian surveillance, and the cruel reality of being forced to train the inefficient systems being positioned to replace them.”
The petition echoes the flyers. It frames the tracking as a potential violation of labor rights. Organizers argue that collective action to improve working conditions receives protection. No opt-out exists for those on company-issued devices. That absence of choice sits at the heart of the protest. Employees aren’t just monitored. They must actively generate the training data. Every click, every hesitation at a dropdown, every keyboard shortcut becomes grist for the model.
This isn’t abstract. Experts have warned for years that white-collar surveillance would expand. Yale law professor Ifeoma Ajunwa noted in the original Reuters reporting that keystroke logging brings to professional workers the real-time oversight long familiar to gig drivers and warehouse staff. No federal U.S. law currently caps such monitoring. In Europe the picture differs. Similar practices would likely breach the General Data Protection Regulation. Some countries ban them outright.
Meta has tried to calm the waters. A follow-up internal communication acknowledged widespread concern. It promised that screen content would be masked during training to prevent memorization. Raw data would face tight access controls. The company insists the effort focuses solely on improving AI agents’ ability to mimic fluid human-computer interaction. Agents that can navigate interfaces without constant human correction promise huge productivity gains. Or so the theory goes.
Yet the backlash reveals a deeper tension. Tech giants have spent years selling AI as a tool that augments human work. Now their own employees experience the other side. They become both the trainers and the potentially displaced. And they don’t like it. Recent coverage from Cybernews highlights how the revolt frames the software as invasive surveillance rolled out amid layoffs. Workers see the contradiction. The same company that built sophisticated global tracking for advertising now applies similar techniques internally. The difference? This time the data helps build replacements.
The protest arrives at a pivotal moment for the industry. AI agents remain imperfect. They struggle with precise interface navigation, with understanding context that seems obvious to humans, with recovering gracefully from mistakes. High-quality demonstration data from actual employees could accelerate progress. But extracting that data without consent or perceived benefit to the workers creates friction. Trust erodes. Morale drops. Some staffers already question whether their daily efforts simply accelerate their own obsolescence.
So far Meta shows no sign of reversing course. The tool continues to run on work devices. Layoffs proceed on schedule. The petition circulates. Organizers in the UK push for union recognition. And those flyers keep appearing in unexpected places. The bathroom stall. The top of the toilet paper dispenser. Places where employees might pause, read, and consider. The question lingers. How many will sign? How far will this pushback spread? One thing seems clear. The era of frictionless AI adoption inside Big Tech has met its first organized resistance. From the inside.
Other outlets picked up the story quickly. Reports from BBC and Fortune detailed the initial rollout last month. They underscored the same core conflict. Companies racing to deploy AI must first train it on human behavior. When those humans work at the company building the AI, the incentives clash. Add mass layoffs and the clash turns combustible.
Industry observers watch closely. If Meta’s employees force concessions here, similar efforts could emerge at other firms pursuing agentic AI. The data demands are real. The human cost feels immediate. Balancing the two will test executive skill in the months ahead. For now, the flyers speak louder than any internal memo. They ask a question no algorithm can yet answer. Do you want to work at the Employee Data Extraction Factory?
WebProNews is an iEntry Publication