Meta pauses AI training program amid investigation into employee data exposure

Meta has paused an internal program designed to collect employee digital activity for artificial intelligence training while it investigates concerns over the exposure of sensitive employee information, as per a Reuters report.The company confirmed the move after internal documents revealed that data gathered through the initiative was accessible to employees across the organisation, raising questions about privacy safeguards and data security practices.Known internally as the Model Capability Initiative (MCI), the program was introduced in April and collected information such as mouse movements, clicks and keystrokes from computers used by employees in the United States. The data was intended to help train and improve Meta's AI models.A Meta spokesperson said the company had built the program with privacy protections and that there was currently no indication that employee data had been improperly accessed. However, the company decided to suspend the initiative while conducting a review."We have carefully designed this program with privacy safeguards and while we have no indication at this time that any data was improperly accessed by Meta employees, we're pausing it while we investigate," spokesperson Tracy Clayton said.Meta did not specify how long the suspension would remain in place. According to a source familiar with the matter, the system was still recording data on Monday afternoon as the pause was being rolled out. The company said the process would take time to implement across all affected employees.The investigation follows the filing of a high-priority internal security incident report, known within the company as an SEV. The report raised concerns that information collected through the program could be viewed more broadly than intended.Internal documentation reviewed by Reuters indicated that exposed information included prompts and transcriptions, private conversations, employee-related performance data and internal sensitivity ratings.The issue has drawn attention because of the nature of the information employees may have accessed through their work devices. One employee participating in the internal discussion called for a deeper investigation into the matter."I have accessed both personal tax and medical information through my work computer, as have many thousands of employees. We were told this data would be protected and only used for valid business purposes after aggressive filtering," the employee wrote, according to the documentation.The latest development comes weeks after reports that the program was collecting more information than employees had initially been told and storing some of the data without encryption, prompting additional concerns about privacy and security controls.Meta has not disclosed how many employees were affected by the issue or whether any changes will be made to the program following the investigation.Also read: Oracle cuts 21,000 jobs in a year, says AI adoption contributed to workforce reduction