BY COMFORT OGBONNA
Meta has suspended an internal employee monitoring program used to help train its artificial intelligence models after discovering potential data security issues involving sensitive staff information.
The social media company announced Monday that it is temporarily halting the initiative, known as the Model Capability Initiative (MCI), while it investigates concerns about how employee data was being stored and accessed within the company. The decision comes after internal reviews revealed that information collected through the program may have been available to a much wider group of employees than originally intended.
Meta said the program was designed with privacy protections and safeguards in place. However, the company decided to pause the initiative out of caution while examining whether internal controls functioned as expected. A company spokesperson said there is currently no evidence that employee data was improperly accessed, but the investigation remains ongoing.
The MCI program was introduced in April as part of Meta’s efforts to improve the performance of its artificial intelligence systems. The tool collects data from U.S.-based employees’ computers, including mouse movements, clicks, and keyboard activity. The information is then used to help train and refine AI models that support the company’s products and services.
Despite the announcement of the suspension, a source familiar with the matter said the system was still collecting data on Monday afternoon. Meta explained that the shutdown process was being rolled out across the organization and could take time before the program was completely disabled for all affected employees.
The investigation intensified after an employee filed a high-priority internal security incident report, commonly known within the company as an SEV. The report raised concerns that sensitive employee information had been exposed through the system.
According to internal documents, the exposed data included full AI prompts and transcriptions, private conversations, employee performance information, and internal data sensitivity ratings. The breadth of the information raised concerns among workers who believed the data would be carefully filtered and protected before being used for AI development purposes.
Questions about the program had already surfaced earlier this year. Reports indicated that the initiative was collecting a broader range of information than many employees initially understood. There were also concerns that some of the data was being stored in an unencrypted format, potentially increasing security and privacy risks.
Internal discussions surrounding the security incident showed growing concern among employees about how personal information may have been handled. One employee called for a deeper investigation into the issue, noting that workers routinely access highly sensitive information through company devices.
The employee wrote that personal tax records, medical information, and other private data had been accessed through work computers by thousands of employees. They said workers were assured that any information gathered through the monitoring program would be adequately protected and used only for legitimate business purposes after undergoing extensive filtering processes.
The incident highlights the growing challenges technology companies face as they seek to collect large amounts of data to train increasingly sophisticated AI systems. As businesses invest heavily in artificial intelligence development, balancing innovation with employee privacy and data security has become an increasingly important issue.
Meta has not provided a timeline for when the investigation will be completed or when the Model Capability Initiative could resume. For now, the company says it is focused on reviewing its internal safeguards and ensuring employee data remains protected while the investigation continues.
Original article
Comments