Company spokesperson Tracy Clayton says, “We have carefully designed this program with privacy safeguards in place and although we have no indication at this time that any data was improperly accessed by Meta employees, we are pausing it while we investigate.”
Meta launched the Model Compatibility Initiative (MCI) tool for US employees in April. According to workers who petitioned against it over privacy, security, and personal liberty concerns, the tool “collects computer inputs such as mouse movements, click locations and keystrokes, as well as screen content.” When MCI launched, employees could not walk out, but this changed to a limited extent after employee protests.
Meta officials have repeatedly defended the data-gathering project, saying it was necessary to train AI systems to operate computer software like humans and that employees were the best examples for artificial intelligence to learn from.
On Monday, a Meta engineer issued an internal security notice saying that databases filled with information collected by MCI had been exposed to anyone inside the company.
A former employee who was actively involved in protesting MCI described the omission as “messed up” – and something that employees expected would happen. “When workers raised concerns, leadership doubled down and failed to acknowledge the risks workers took regarding the security and privacy of employee and customer data,” the person says. “The leadership has clearly created an authoritarian environment where workers are no longer respected or listened to.”
But after critical comments appeared on internal forums on Monday expressing frustration over the security issue, Meta surprised some of its employees by halting MCI altogether, telling WIRED about the development several hours before it was announced to employees. Some employees told WIRED that they were confused in the meantime because the tool was constantly running on their laptops.
Late Monday night, Stephen Kasriel, Meta’s vice president who oversees AI research, announced the pause and told employees that the security issue was discovered on June 18 and resolved within four hours. But the initial solution could not be found and access to the data had to be further closed. The issue has made “some MCI-derived data” accessible to more people than expected, he wrote, without elaborating.
“We will re-enable MCI only when we are confident in the effectiveness of our data security controls,” Casreal said. He said Meta will share more information about the future of MCI now that it has “collected enough data to assess the long-term value of the tool.”
Zoë Schieffer contributed to this report.
Updated 12:15am EDT, 6/23/2026: This story was updated with details from Stephen Kasriel’s staff note and additional comments from the staff.
<a href