Sign up to see the future, today
Can’t-miss innovations from the bleeding edge of science and tech
Meta’s controversial program that spied on its employees’ computers has backfired spectacularly.
Though its chief technology officer Andrew Bosworth assured that the data it collected would be “tightly controlled,” the company is now pausing the tracking program after sensitive employee information was leaked internally, according to reports from Business Insider and Wired.
A security notice sent out Monday said that the exposed data included employee’s full AI prompts and transcriptions, performance data, and even private conversations. The leak allowed the data to be accessible to any employee inside the company.
Meta said it’s investigating the incident and confirmed it paused the program, but maintains it’s keeping a lid on things.
“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,” a spokesperson told outlets.
The program, dubbed the Model Capability Initiative, was intended to gather data so Meta’s AI models could learn “how people actually complete everyday tasks using computers.” Reports suggested it collected everything from employee keystrokes to recording their computer screens.
When it was announced in April, it immediately sparked backlash in the workforce, with internal posts openly denouncing the initiative as an invasion of privacy, and some employees circulating a petition calling to end it. It came while morale at the company was at a nadir, following fresh layoffs that forced out nearly 8,000 employees, and a heavy push from the top that workers should heavily use AI to produce as much code as possible.
Under CEO Mark Zuckerberg’s directive, the company has doubled-down even further on developing automation tech, moving employees who were working elsewhere onto a new AI project, sowing further discontent.
Reactions from employees about the leak were unequivocally critical.
“I am incensed,” one employee wrote Monday in an internal group, per a screenshot obtained by BI.
“I don’t see any evidence of malicious access, but the fact that this data wasn’t locked down as originally promised is super frustrating,” another complained.
Memes also abounded. In an internal forum, one employee posted an image of Jim Halpert from “The Office” holding a sign that says, “0 days since our last nonsense.”
Bosworth, Meta’s CTO who promised that the collected data would be “tightly controlled,” responded to employees’ posts by admitting that the initiative’s implementation had fallen short of the standards set in its prevacy review, per Wired.
The company, however, is poised to resume the data tracking program once the issue is settled.
“We will only re-enable MCI when we are confident in the effectiveness of our data protection controls,” Stephane Kasriel, a Meta vice president overseeing AI research, told employees Monday, according to Wired. Kasriel said the company discovered and resolved the issue last week, but that the initial fix didn’t work.
It’s not the only recent security blunder at Meta. In March, a rogue AI agent gave an employee advice that it wasn’t authorized to share, resulting in sensitive data being exposed and causing what was described as a critical security incident.
More on Meta: Meta Exec Admits Zuckerberg Has Crushed Workers’ Spirits
