On June 22, 2026, Meta made the decision to suspend its AI training endeavor, known as the 'Model Capability Initiative (MCI).' This initiative had been gathering data by monitoring employees' digital behaviors, including mouse movements and keyboard inputs, with the aim of developing AI agents proficient in independently executing office-related tasks. The suspension came in the wake of an internal data breach: confidential data amassed via MCI, encompassing employees' personal chats, work performance metrics, and voice-to-text transcriptions, was unintentionally exposed to the entire workforce. A Meta spokesperson acknowledged that an inquiry is in progress but refrained from specifying the length of the suspension. Earlier, Meta had already curtailed the scope of the program in response to employee dissent and regulatory scrutiny from the EU, introducing a mechanism to halt data acquisition. This episode underscores once more the privacy and security hurdles that tech firms encounter as they strive to advance their AI ambitions.