Meta Pauses Internal AI Training Program After Data Leak Meta has halted an internal AI training program that monitored employee activity after a leak exposed how widely sensitive staff data could be accessed inside the company, intensifying long‑running concerns over privacy and security.

Early rollout and employee backlash

In April, Meta introduced the Model Capability Initiative (MCI), an AI training effort that used employees’ keystrokes and mouse movements as data to improve company models. The program was reportedly mandatory for most staff and quickly drew internal criticism from workers uneasy about having their activity recorded and analyzed.

Leak exposes scope of accessible data

Screenshots later obtained by Business Insider showed that data from the program — including employees’ private conversations, performance details, and transcriptions — could be viewed across the entire company. The incident was classified internally as a “SEV 2” on Meta’s 0–5 severity scale, indicating a serious but not top‑level emergency.

Inside Meta, frustration mounted as workers realized the data had not been locked down as they say was originally promised. One employee wrote, “I am incensed” and called it “super frustrating” that protections were not in place from the start, according to internal messages seen by Business Insider.

Meta pauses program and opens investigation

Following the leak, Meta paused the MCI program, which had already been controversial due to its tracking capabilities. In a statement, the company said it had “carefully designed this program with privacy safeguards” and that there was “no indication at this time that any data was improperly accessed by Meta employees,” but added that it would keep the initiative on hold while an investigation proceeds.

The episode is the latest in a series of security issues involving Meta’s AI systems, including a recent flaw that allowed its chatbot to help hijack Instagram accounts and a separate incident involving a rogue AI agent. Together, they underscore a growing tension between Meta’s aggressive AI ambitions and internal expectations for data protection.

Continue reading https://foxvector.com/stories/019efc9e-8e18-11a8-724d-3815d962f0b1