Meta Exposed Data Internally From Its Controversial Employee-Tracking Program

Meta inadvertently exposed employee keystroke, mouse-click, and screen-content data collected for AI training to anyone inside the company, according to an internal security notice and three current employees [1]. The data was gathered under the Model Capability Initiative, a program that began monitoring US employees' corporate laptops in April [1]. The security notice, sent Monday, indicated that "employee data across 45,000 hive tables" had been exposed, including "full prompts and transcriptions, private conversations, people and performance data" [1]. Meta spokesperson Tracy Clayton confirmed the company is investigating. "We have carefully designed this program with privacy safeguards," Clayton said, adding, "we have no indication at this time that any data was improperly accessed by Meta employees" [1]. The incident adds to a long record of privacy scrutiny facing the company. Meta's revenue model, which relies on collecting user information for advertising, has drawn criticism from regulators and advocacy groups [4]. In September 2024, the Federal Trade Commission released a report finding that data practices at major platforms, including Meta, put individuals at risk of identity theft, stalking, and reputational harm [4]. Inside Meta, the exposure validated concerns employees had raised since the tracking began. Last month, more than 1,600 employees signed an internal petition warning that "collecting this data introduces both security and regulatory risks for Meta, including the potential for breaches and unauthorized disclosure" [1]. One engineer circulated a note calling the screen-scraping an invasion of privacy [1]. Chief Technology Officer Andrew Bosworth acknowledged in an internal post that the program's implementation had fallen short of the standards outlined in its privacy review [1]. Bosworth had separately apologized for the company's "atrocious" communication around an AI reorganization that moved roughly 6,500 employees into new roles focused on improving AI models [1]. Meta has begun offering more exemptions to the monitoring, including allowing staffers to briefly turn off the surveillance for sensitive tasks such as scheduling personal appointments [1]. Some employees continue to demand the tracking be halted entirely [1]. The company's push into AI research spans multiple fronts. Meta AI, its dedicated research division, develops artificial intelligence and augmented reality technologies [11]. External researchers have also benchmarked Meta's open-weight language models, including Llama 4 Scout, for tone sensitivity in prompt engineering, finding that rude prompts can reduce accuracy in certain humanities tasks [7]. CEO Mark Zuckerberg defended the data-gathering project in a leaked company meeting, stating that "AI models learn from watching really smart people do things" and that the "average intelligence of the people who are at this company is significantly higher" than outside contractors [1]. Sources told WIRED the incident has been marked as closed, suggesting it was likely resolved [1].