Meta has suspended an internal programme designed to collect data on how employees use their computers, following concerns from workers over privacy and the potential use of the data to automate jobs.
The programme, called the Model Capability Initiative (MCI), was introduced by Meta’s artificial intelligence unit TBD to generate data that could help train AI systems to perform everyday computer-based tasks. The initiative involved collecting information such as keystrokes, mouse movements and screen activity from employees.
Meta’s AI teams viewed employee-generated computer-use data as a way to address gaps in AI systems that can perform specialised tasks such as programming and research but have struggled with routine computer operations.
The initiative faced opposition after its internal launch in April. Employees raised concerns that the system could capture sensitive information, including passwords, immigration documents and financial details. There were also concerns that the data could eventually be used to automate work currently carried out by employees.
Opposition included internal campaigns, flyers and a petition signed by more than 1,800 Meta workers. Employees also objected to the limited opt-out mechanism introduced in June, which allowed them to disable the system temporarily for personal activities.
The programme faced a further setback in June after employees discovered that data collected through the software could be accessed across Meta and had included private conversations. Meta subsequently suspended MCI and said any future version would be made opt-in.
The programme remained suspended two months later. There was also uncertainty over whether the data collected had been used to train Meta’s latest AI models, with differing internal assessments of its use and value.
