Meta Suspends AI Training Tool After Internal Data Breach Exposes Staff Activity
The social media giant claims no evidence of improper access, but internal dissent intensifies over privacy safeguards and the ethics of surveillance for artificial intelligence development.
Meta has suspended its Model Compatibility Initiative (MCI), a controversial internal programme designed to track employee computer activity, following an internal security breach that exposed collected data to other staff members. The company stated it is investigating the incident, though it claims there is no evidence of improper access by employees. The tool, rolled out in April to US workers, faced significant internal opposition regarding privacy and security concerns before being suspended.
The MCI tool collects granular computer inputs, including mouse movements, click locations, and keystrokes, as well as screen content. According to an internal security notice, the exposed tables contained employee activity such as full prompts, transcriptions, private conversations, people data, and performance metrics. The incident involved data accessible across 45,000 internal database tables, raising immediate concerns about the scope of the data exposure.
Meta executives have defended the data-gathering project, stating it was necessary to train AI systems to operate computer software the way humans do. CEO Mark Zuckerberg has argued in a leaked audio recording that AI models learn best by watching high-performing employees. However, the initiative began in April and has been a source of significant internal dissent, with some engineers describing the scraping of their laptop screens as an invasion of privacy and exploitation.
When MCI first launched, employees could not opt out, but this changed to a limited degree after workers protested. A former employee involved in pushing back against MCI described the lapse as "a mess" and one that employees had expected would occur. This individual criticised leadership for doubling down on the project despite raised concerns about the safety and privacy of worker and customer data, characterising the environment as one where workers are no longer respected or heard.
On Monday, critical comments poured into internal forums regarding the security issue, shocking some staff when Meta paused the programme. The company informed WIRED of the development before announcing the suspension to employees. Meta spokesperson Tracy Clayton stated that the company has no indication at this time that any data was improperly accessed by Meta employees, adding that the programme was carefully designed with privacy safeguards. The investigation into the extent of the exposure and potential misuse remains ongoing.
