More than 1,600 Meta employees have signed an online petition demanding their company scrap a program that logs their keystrokes, mouse movements and screen interactions to train artificial intelligence models. The effort, hosted at mcipetition.com, represents one of the most visible internal challenges yet to Silicon Valley’s rush to feed human behavior into AI systems.
The initiative in question carries the name Model Capability Initiative, or MCI. Launched quietly in recent months, it installs software on U.S.-based employees’ work laptops. That software records how people navigate applications, where they click, what they type and even periodic snapshots of screen content. The stated purpose is straightforward. Meta wants its AI agents to learn realistic patterns of computer use so the systems can one day handle tasks on their own.
But many inside the company see something different. They see nonconsensual extraction of personal data. They see a troubling precedent for how far employers will go to fuel AI development. And they worry the collected information could expose sensitive details ranging from Social Security numbers to confidential business information.
The petition’s language pulls 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 states. Organizers add that “AI meant to serve and empower people cannot do so while disregarding or disrespecting the consent of the people it claims to serve.” They demand Mark Zuckerberg and Meta’s leadership team kill the program entirely, immediately and forever.
This pushback did not emerge in a vacuum. Reuters first reported on the tracking software in April, detailing how it captured activity across tools including Google, LinkedIn, Wikipedia, GitHub, Slack and internal Meta systems. (Reuters, April 2026)
Employees quickly took notice. Internal posts expressed alarm. One software engineer told colleagues he did not want his screen scraped because it felt like an invasion of privacy. Zooming out, he wrote, he did not want to live in a world where humans are exploited for training data. That post gained thousands of likes. Flyers soon appeared in Meta offices across California and New York, taped to vending machines, bathroom stalls and meeting rooms. They labeled the company an “Employee Data Extraction Factory” and directed workers to the petition. (Reuters, May 2026)
The timing sharpened the tension. Meta had already signaled plans to cut roughly 10 percent of its workforce as it reoriented toward AI. Thousands of engineers and product managers found themselves reassigned to teams focused on training generative models. Some joked in internal chats that they would rather face layoffs than the new data-labeling duties. Morale, already strained, dropped further.
Meta responded in early June. The company rolled out adjustments. Employees could now pause data collection for up to 30 minutes at a time. They could request exemptions. The team behind MCI also introduced battery-life optimizations after complaints that the software drained laptop power and spiked home internet usage. Stephane Kasriel, a vice president in Meta’s Superintelligence Labs, outlined the changes in an internal memo. (BBC, June 2026)
Yet petition organizers were not satisfied. Partial mitigations and selective opt-outs for executives fell short of their core demand. They want the entire effort abandoned. As of early June the signature count had climbed past 1,600, drawn from full-time employees, contractors and interns. Names appear in random order on the site after validation to confirm Meta affiliation. The petition notes protection under the National Labor Relations Act, which shields concerted activity over working conditions from retaliation.
Concerns extend beyond privacy. Past incidents at Meta give signers pause. The company paid a €91 million fine in 2024 after Irish regulators found plain-text passwords stored in a database accessible to employees, according to a BBC report. More recently, in March 2026, a Meta AI agent reportedly provided incorrect instructions that led to a large leak of sensitive user data to other employees. The Guardian covered that episode. (The Guardian, March 2026)
California privacy law adds another layer. Under the California Consumer Privacy Act and its successor, workers have rights to know what data is collected about them, to request deletion and to limit certain uses. Petition supporters argue that MCI runs afoul of those expectations when it sweeps up behavioral patterns without genuine consent.
The debate reflects deeper questions now confronting the technology industry. As companies race to build more capable AI agents, they need rich examples of how humans actually work. Corporate email, spreadsheets, code editors and browsers contain that data in abundance. But whose data? And at what cost?
Meta has argued internally that the tracking is not for performance evaluation. It exists solely to improve AI. In an early statement to the BBC, the company said that if it is building agents to help people complete everyday tasks on computers, its models need real examples of how people actually use them. (BBC, June 2026)
Critics counter that good intentions do not erase the power imbalance. Employees cannot easily say no when the software arrives pre-installed on company machines. Executives enjoy opt-outs that appear unavailable to everyone else. And once data enters training pipelines, control over its future use becomes difficult to guarantee.
Similar frictions have surfaced elsewhere. Google DeepMind workers in Britain voted to unionize partly over concerns about military AI applications. Laid-off Oracle employees signed letters protesting what they described as use of their labor to train AI before their dismissal. The pattern suggests growing resistance among knowledge workers who see their skills and behaviors being repurposed to accelerate their own potential replacement.
At Meta the discontent has taken creative forms. One departing staffer released a protest video that name-checked Chief Technology Officer Andrew Bosworth and urged colleagues to sign the petition. Set to music, it included the line “Sign the petition, no more wishing, just deny MCI.” (Wired, May 2026)
Whether the campaign produces lasting change remains uncertain. Meta has not retreated from its broader AI ambitions. Spending on the technology is projected to reach between $125 billion and $145 billion this year, more than double the previous year’s outlay. CEO Mark Zuckerberg has tied the company’s future to progress in generative systems and autonomous agents.
But the petition has forced a conversation. It has prompted adjustments to the program. It has highlighted the gap between Meta’s public commitments to build AI responsibly and the methods used to gather training data. The company’s own code of conduct, quoted in the petition, calls on leaders and contributors alike to hold themselves to a high standard and speak up when needed.
Organizers have taken that charge seriously. “We are speaking up,” they write, “because it’s expected of us and, more importantly, because it is the right thing to do.”
So far Meta has offered no public comment beyond the internal memo on adjustments. A spokesperson declined requests from multiple outlets. The petition site itself discourages direct media links in an apparent effort to limit external visibility while the internal campaign continues.
Yet the story has already spread. Coverage in Reuters, the BBC, Wired and others has amplified employee concerns beyond the company’s walls. Recent analyses connect the MCI controversy to broader workforce anxiety as AI reshapes job responsibilities across Silicon Valley. One report noted that morale at Meta is plummeting amid forced AI-related reassignments and surveillance tools.
For industry observers the episode offers a rare window into tensions that usually remain behind corporate firewalls. Tech giants have long collected user data at scale. Now they are turning the same gaze on their own workers. The question is whether employees will tolerate it or whether this rebellion marks the start of more organized resistance to the data demands of the AI age.
Meta’s next moves will matter. If it doubles down despite the outcry, it risks further eroding trust inside its ranks. If it scales back more substantially, it may signal that even the most powerful companies cannot ignore their own people when those people push back in coordinated fashion.
Either outcome will shape how other firms approach similar initiatives. The petition at mcipetition.com has already collected signatures from a significant slice of Meta’s workforce. Its organizers show no sign of stopping. And in the current climate of rapid AI development, that alone makes it a development worth watching closely.
WebProNews is an iEntry Publication