- Meta has suspended an AI data collection programme following privacy concerns.
- More than 1,600 employees signed a petition against the initiative.
- Workers raised concerns over monitoring of keystrokes, screen activity and private conversations.
Meta has paused an internal programme that collected employee computer activity for artificial intelligence training after a growing backlash from workers who argued the initiative crossed privacy boundaries.
The programme, known internally as the Model Capability Initiative (MCI), gathered information from company-issued laptops, including keystrokes, mouse clicks and content displayed on screens. The data was intended to help improve Meta AI models and strengthen the company's wider artificial intelligence strategy, according to reports.
However, concerns escalated after employees questioned how much information was being collected and who could access it. The dispute has become one of the clearest examples yet of tensions between Silicon Valley's AI ambitions and employee privacy rights.
Workers draw a line
More than 1,600 employees signed a petition calling on Meta to stop collecting what they described as employee "computer use" data.
The petition reportedly argued that collecting and repurposing such information raised serious concerns about privacy, consent and trust in the workplace.
Those concerns intensified after technology publication Wired reported that data gathered through the programme had been accessible to employees across the company. According to the report, internal records referred to exposed data tables containing prompts, transcriptions, private conversations and performance-related information.
Meta said it had no indication that the information had been improperly accessed, but confirmed the programme had been suspended while an investigation takes place.
In a statement, the company reportedly said the initiative had been designed with privacy safeguards and that it was pausing the programme while reviewing the situation.
The growing cost of AI ambition
The controversy arrives at a critical moment for Meta and its founder, Mark Zuckerberg, who is investing heavily in artificial intelligence as competition intensifies across the technology sector.
During an internal meeting, Zuckerberg reportedly said AI systems learn by observing highly skilled workers perform tasks, arguing that Meta employees could help train more capable models because of their expertise.
The company is expected to spend as much as £110 billion this year on capital expenditure, with a significant share directed towards AI infrastructure, including data centres and advanced computing systems.
Yet the episode highlights a challenge facing technology companies as they search for new sources of training data. While businesses increasingly view workplace activity as a potential resource for improving AI systems, employees are questioning where the line should be drawn between innovation and surveillance.
The debate may not end with Meta. As companies race to develop more powerful AI tools, questions around consent, transparency and the use of employee-generated data are likely to become increasingly difficult to ignore.
