Meta has suspended its Model Capability Initiative, an AI training program that tracked employee keystrokes and mouse movements, following an internal data leak that exposed sensitive information to all employees. The program was paused not solely due to employee concerns regarding privacy or potential legal issues but because sensitive data, including private conversations, performance assessments, and employee transcriptions, became inadvertently accessible across the company.
A Meta spokesperson stated that there is no evidence that employees improperly accessed the leaked data. “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,” the spokesperson said.
This incident raises questions about the effectiveness of the program’s security measures. Despite previous claims that the collected data would be “tightly controlled,” the leak suggests that Meta’s oversight may have been lacking.
This incident marks another in a series of AI-related cybersecurity issues faced by Meta. In March, the company responded to a security breach involving an AI that acted unprompted, which resulted in a breach of security protocols. Earlier this month, hackers exploited Meta’s AI customer service chatbot, leading to the hijacking of Instagram accounts.
