Meta has temporarily suspended its internal AI training programme after an incident reportedly exposed sensitive employee information across the organisation, raising fresh concerns around workplace data privacy and AI-driven monitoring practices.
According to reports, the issue allowed access to employee-related information, including private conversations, performance data and meeting transcriptions. The incident was internally classified as a Severity 2 (SEV 2) event, indicating a significant operational concern that required immediate attention.
Meta has confirmed the incident and launched an investigation into the matter. The company said the programme had been designed with privacy safeguards and that there is currently no indication that employee data was improperly accessed. However, it has decided to pause the initiative while the review is underway.
The programme, known as the Model Capability Initiative (MCI), was introduced earlier this year to help improve Meta’s AI models by analysing workplace interactions such as keystrokes and mouse movements. Participation was reportedly mandatory for most employees, prompting concerns among staff about the extent of workplace monitoring and the collection of behavioural data.
The latest development has intensified scrutiny of the initiative, particularly as organisations increasingly turn to employee-generated data to train AI systems. Employees have reportedly expressed frustration that stronger access controls were not implemented from the outset, especially given the sensitive nature of the information involved.
The incident highlights the growing challenge companies face in balancing AI innovation with employee trust and data protection. As businesses accelerate AI adoption, workplace surveillance, consent and privacy governance are emerging as critical HR and organisational issues.
