Meta's Creepy Model Capability Program Unmasked as Security Glitch Exposes 1000s of Workers' Private Messages

Meta has been forced into damage-control mode after an internal employee-monitoring program reportedly exposed sensitive staff data across the company, triggering alarm among workers and fresh scrutiny of the tech giant's workplace surveillance practices.

The incident has sparked internal backlash and raised uncomfortable questions about how the company handles employee privacy while pursuing its ambitious artificial intelligence goals.

Meta's Employee Monitoring Program Unravels

A highly controversial internal program designed by Meta to monitor workers' laptops has collapsed into a major corporate disaster. Despite assurances from Chief Technology Officer Andrew Bosworth that the collected metrics would remain 'tightly controlled', the technology giant has been forced to suspend the operation.

The sudden freeze follows an internal security blunder that inadvertently leaked confidential employee records across the firm, according to leaked images viewed by Business Insider and an internal security notice seen by WIRED.

A security alert issued on Monday revealed that the exposed material contained full AI queries, audio transcripts, job reviews and private messages. The blunder meant that anyone working at the company could easily view the highly sensitive information.

While Meta confirmed it has suspended the program while it investigates the matter, the tech giant insists it has the situation under control. Spokesperson Tracy Clayton reiterated that the firm is looking into the security flaw, stating: 'We have carefully designed this program with privacy safeguards,' he said, adding: 'we have no indication at this time that any data was improperly accessed by Meta employees.'

45,000 Databases Caught Up in Leak

According to Monday's security alert, the breach compromised 'employee data across 45,000 hive tables'. The affected databases held detailed records of worker activity, including 'full prompts and transcriptions, private conversations, people and performance data'.

For many Meta workers, the security disaster was the ultimate 'I told you so' moment. Internal message boards quickly lit up with staff pointing out that this was exactly the kind of nightmare they had warned about back in April, when the company first began monitoring their laptops under the so-called Model Capability Initiative.

Staffers did not hold back on internal forums following Monday's leak. Frustrated employees demanded to know how Meta's privacy checks could fail so catastrophically, while others asked whether everyone affected would be invited to the upcoming briefing on the disaster.

Employees Vent Anger After Privacy Failure

Meanwhile, the mood turned cynical on a company message board popular for workplace humour, where one employee took a jab at the situation by posting a meme of The Office character Jim Halpert holding a sign that reads: '0 days since our last nonsense.'

The workforce was united in its fury over the security failure.

'I am incensed,' one worker posted in an internal group on Monday. Another voiced their frustration online, writing: '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.'

Facing the heat on internal forums, Chief Technology Officer Andrew Bosworth — the man who originally swore the gathered information would remain 'tightly controlled' — stepped in to admit the rollout had completely failed to meet the standards set by the company's own privacy review. Yet Meta is not throwing in the towel just yet; the tech giant intends to restart the tracking system once the current issues have been resolved.

Meta Plans to Revive Suspended System

Meta's vice president of AI research, Stephane Kasriel, tried to reassure staff on Monday by stating: 'We will only re-enable MCI when we are confident in the effectiveness of our data protection controls.'

Kasriel explained that although the tech giant had spotted the flaw and rolled out a fix the previous week, that initial patch unfortunately failed to stick.

How Meta's AI Training System Worked

The tech giant launched the Model Capability Initiative to train its AI systems by watching 'how people actually complete everyday tasks using computers'.

To feed those algorithms, the operation reportedly gathered an extraordinarily invasive amount of data, capturing every keystroke workers made and constantly recording their computer screens.

The program sparked an immediate revolt among staff when it debuted in April, triggering fierce internal posts that condemned the scheme as a massive privacy violation, alongside a worker petition demanding its termination.

The timing could not have been worse for workforce morale, which had already hit rock bottom after a brutal wave of layoffs eliminated nearly 8,000 jobs, all while executives aggressively pressured the remaining staff to rely heavily on AI to churn out code.

Chief Executive Mark Zuckerberg's relentless push into automation has only fuelled the flames of workforce resentment. Under his orders, Meta has aggressively shifted staff away from their original roles and on to a new artificial intelligence project, a top-down shake-up that has triggered deep frustration across the company.