Meta installed software on thousands of U.S. employee laptops to record every mouse movement, keystroke and screen interaction. The goal? Gather real-world examples to train AI agents that mimic human computer use. Workers responded with flyers in bathrooms, petitions and pointed internal posts. The backlash reveals deep tensions at a company racing to build artificial intelligence while shrinking its workforce.
The program, known internally as the Model Capability Initiative or MCI, began rolling out in April. It captures clicks, button navigation, dropdown selections and periodic screen snapshots when employees work in approved applications such as Gmail, chat tools or code editors. Meta told staff the data would stay limited to training purposes with safeguards in place. Many didn’t buy it.
“Selfishly, I don’t want my screen scraped because it feels like an invasion of my privacy,” one engineer wrote in an internal post viewed by nearly 20,000 colleagues. “But zooming out, I don’t want to live in a world where humans—employees or otherwise—are exploited for their training data.” The message, reported by WIRED, mixed personal discomfort with broader alarm about norms around data extraction.
That post didn’t stand alone. On Meta’s internal communications platform, the top comment under the official rollout announcement read simply, “This makes me super uncomfortable. How do we opt out?” Angry emojis dominated the reactions, according to documents obtained by Business Insider.
By mid-May the discontent turned visible. Employees plastered flyers across multiple U.S. offices. They appeared in meeting rooms, on vending machines and even atop toilet paper dispensers. The pamphlets carried a blunt question: “Don’t want to work at the Employee Data Extraction Factory?” They urged colleagues to sign an online petition citing protections under the National Labor Relations Act for organizing around working conditions. The action came roughly one week before the first wave of planned layoffs affecting about 10% of the workforce, as detailed in a Reuters investigation.
Protests surfaced just as Meta prepared major job cuts tied to its AI strategy.
Some workers saw direct connections. They worried that the very data collected from their daily tasks would help build systems capable of replacing parts of their roles. One union representative captured the sentiment sharply. “Meta’s workers are paying the price for management’s reckless and expensive bets,” Eleanor Payne of United Tech and Allied Workers told Reuters. “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 itself pulled no punches. “It should not be the norm that companies of any size are permitted to exploit their employees by nonconsensually extracting their data for the purposes of AI training,” it stated, per the WIRED report. Organizers framed the effort as a stand against setting dangerous precedents. And some staff delayed installing the software, tolerating repeated pop-up notifications as a form of quiet protest.
Meta defended the initiative. Spokesperson Andy Stone explained the need for authentic data. “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.” Company statements emphasized that the tool operated only on U.S.-based machines and included privacy protections. Spokesperson Dave Arnold, quoted in Mashable, added that Meta notified non-U.S. employees about potential data capture in emails or chats with U.S. colleagues “in the interest of transparency.” He stressed the firm had “carefully considered and mitigated potential privacy risks” and remained “committed to complying with applicable laws and regulations.”
Yet questions persisted. Internal complaints described the tool reaching further than initially described. It reportedly logged code changes, computer sleep and wake cycles, and even URLs copied to the clipboard. Data usage reportedly spiked some employees’ home internet consumption, exhausting monthly quotas in days. One internal post raising these issues vanished. Meta called it “fundamentally inaccurate,” according to two employees who spoke with Mashable.
Documents reviewed by Reuters revealed another wrinkle. MCI could capture the contents of any email or direct message sent to a U.S. employee, no matter where the sender worked. A legal expert consulted by the publication suggested this practice might clash with Europe’s General Data Protection Regulation. The issues centered on whether collecting such data counts as incidental and whether it satisfies purpose limitation requirements. A related Reuters report from late May highlighted the tool’s potential collision course with EU privacy rules, though full details remain under discussion inside the company and among regulators.
The episode fits a larger pattern. Meta has tied much of its recent cost-cutting to artificial intelligence priorities. Earlier layoffs, including thousands in previous rounds, carried similar justifications. Remaining employees described record-low morale in conversations with WIRED. Sixteen current and former staffers pointed to the surveillance push as a leading driver of frustration and, in the UK, renewed union activity. British workers launched organizing drives with the United Tech and Allied Workers, even creating a site called Leanin.uk in a wry nod to former executive Sheryl Sandberg’s book.
But the protests also highlight contradictions inside Big Tech. Companies push AI as the future while asking humans to generate the training data that powers it. Employees who once viewed their work as creative or strategic now see parts of it reduced to labeled interactions. The engineer whose post gained traction captured this unease. “I’m mixed on AI. On one hand, I really enjoy using it to write software. On the other hand, I’m really nervous about its impact on the world.” The writer went further, linking the tool to a gradual degradation in company culture over more than a decade, accelerating in recent years amid efficiency drives and budget pressure.
Meta has not disclosed whether the collected data has measurably improved its models. Nor has it commented directly on the petition or flyer campaign beyond removing some posted materials. Company representatives declined to address specific allegations when asked by WIRED. The silence leaves workers to weigh personal privacy against corporate ambitions in an environment where monitoring latitude for U.S. employers remains broad.
Petition organizers emphasized collective voice. “Your voice matters,” the viral internal post concluded. “Moments like this are why I was drawn to Facebook in the first place.” That appeal resonated. Flyers stayed up in some locations despite removal efforts. The online petition gathered signatures. UK union efforts gained momentum. And internal chatter continued.
What happens next remains unclear. The layoffs proceeded. The software stays installed. Yet the visible pushback marks a rare public fracture in a workplace long known for its intensity and alignment with leadership goals. It signals that even at Meta, the drive to feed AI with human behavior has limits. Employees are drawing them. Whether the company adjusts course or simply absorbs the dissent will shape its internal culture for years ahead. The data keeps flowing. The questions about its true cost linger.
WebProNews is an iEntry Publication