Recently, Meta employees distributed flyers in multiple U.S. offices on Tuesday to protest the company's recent installation of mouse-tracking software on their computers. Photos of the flyers seen by Reuters show this activity. The flyers appeared in conference rooms, above vending machines, and even on toilet paper holders in the offices of the Facebook parent company, urging employees to sign an online petition against the move.
According to photos seen by Reuters, the flyers read: "Don't want to work in an 'Employee Data Extraction Factory'?"
This flyer distribution occurred about a week before Meta plans to lay off 10% of its workforce (approximately 8,000 out of 78,865 employees), with further cuts planned for the second half of 2026. According to Trueup data, the tech industry has already eliminated over 95,000 positions across 247 layoff events in 2026, averaging 882 jobs lost per day. Against this backdrop, Meta installed software on employee computers capable of recording mouse movements, clicks, and activity paths.
This is one of the clearest signals yet that a labor movement is gradually taking shape within this Silicon Valley giant: some employees are beginning to channel their anger over the company's plan to restructure its workforce around AI into efforts to organize labor action. And the pressure driving this movement is not unique to Meta.
This Protest is Organized and Legal
This protest is not a spontaneous outburst but a somewhat organized action. It is understood that the flyers they distributed and the related online petition cite the U.S. National Labor Relations Act (NLRA), reminding signers that employees "are legally protected" when they choose to improve their working conditions through organized action.
It is worth noting that the citation of the U.S. NLRA in the protest flyers is not merely rhetorical decoration but a clear legal signal that human resource managers need to take seriously. The U.S. National Labor Relations Board (NLRB) has clearly stated that using AI to interfere with employees' organizational rights is illegal, especially when it involves data collection or employee monitoring. This statement places "data-collecting mouse-tracking software used to train AI models" in a legally sensitive area, particularly when the company is simultaneously conducting 10% layoffs.
According to foreign media reports, the NLRB previously ruled that Meta's confidentiality agreements were unlawful, finding that clauses prohibiting laid-off employees from discussing working conditions infringed upon employees' organizational rights. The current protest activity where employees publicly disseminate company monitoring information is precisely the type of activity the NLRA aims to protect.
While the flyers guide employees to participate in the petition, in the UK, a group of Meta employees have also partnered with United Tech and Allied Workers (UTAW, affiliated with the Communication Workers Union) to launch a formal unionization campaign. These employees have also set up a website, recruiting members through a specific URL that pays homage to former Chief Operating Officer Sheryl Sandberg's bestselling book Lean In, which encourages women to pursue equality in the workplace.
A UTAW representative confirmed this action. UTAW organizer Eleanor Payne stated, "Meta employees are paying the price for management's reckless and expensive bets. While executives chase speculative AI strategies, employees face devastating layoffs, intrusive surveillance, and the brutal reality of being forced to train inefficient systems that may ultimately replace them."
Compared to Meta's overall employee size, this action remains small but touches on an issue of "internal cohesion" the company has rarely faced before. The company's last significant employee protest was a collective walkout in 2018 surrounding sexual harassment policies, which ultimately ended with policy adjustments rather than a crackdown on employees.
Meta Defends Itself: Models Need Real Examples
During an earnings call in January this year, Meta CEO Mark Zuckerberg stated that 2026 will be "the year AI begins to fundamentally change how we work." Last month, Meta notified employees about the launch of the "Model Capability Initiative" (MCI), which captures information such as employee mouse clicks, keyboard inputs, and on-screen content context, then uses the collected data to train AI agents.
According to an internal memo seen by Reuters, the "Model Capability Initiative" (MCI) runs on company-issued devices. Meta describes it as "spiritually voluntary" but effectively mandatory for employees using designated applications. Whether this practice can withstand scrutiny in jurisdictions with stricter employee privacy protections remains unclear; in contrast, current EU workplace monitoring rules set higher thresholds than U.S. federal law regarding the "principle of proportionality" and "employee consent."
From a purely technical perspective, the dataset MCI aims to generate does hold value for certain AI training paradigms. Machine learning models typically benefit from real human-computer interaction data to achieve nuanced performance. The idea is to create artificial intelligence that can learn from observed human behavior, similar to how junior employees learn by observing seniors. However, the root of the ethical and practical issues lies in the data collection mechanism. Crucially, Meta has not yet made public MCI's API, configuration keys, or version number. This lack of transparency makes independent auditing of the software's specific functions and limitations difficult, fueling employee suspicion.
Now, within the company, Zuckerberg's statement has been interpreted by some employees as identifying which roles are being "incorporated into the dataset." "This makes me extremely uncomfortable," wrote one engineering manager on an internal message board. Others worry they are helping to train systems that may replace them in the future. "How do we opt out?" one employee asked. According to foreign media reports, Meta Chief Technology Officer Andrew Bosworth reportedly confirmed that they effectively cannot opt out.
For months, Meta employees have been voicing dissatisfaction on internal platforms and online forums regarding the company's plan for large-scale layoffs this year (which was confirmed to employees over a month after it was first reported), as well as the introduction of mouse-tracking software. The tracking program records mouse movements, clicks, keystrokes, and screenshots within a designated list of work applications. Many employees believe this is tantamount to helping design robots that will replace them.
When asked about this, Meta spokesperson Andy Stone offered a relatively straightforward business explanation: "If we want to build AI agents that can help people use computers to perform everyday tasks, our models need real-world usage examples, such as mouse movements, button clicks, and navigating drop-down menus." Meta also stated in a declaration that this data is used to teach AI agents how to operate software, and it only runs on specified applications and websites, not across all computer activity. Furthermore, they have implemented "safeguards" to protect company-sensitive information.
As for how many more employees will lose their jobs, Meta is still evaluating. Meta CFO Susan Li told investors in April, "We actually don't know yet what the optimal size of the company will be in the future. I think there is just a lot of change right now, especially against the backdrop of rapidly advancing AI capabilities."
Reference Links:
https://www.reuters.com/sustainability/society-equity/meta-us-employees-organize-protest-against-mouse-tracking-tech-2026-05-12/
https://www.engadget.com/2172212/meta-employees-are-protesting-the-companys-mouse-tracking-program/
This article is from the WeChat public account "AI Frontline," compiled by Huawei.
