Revolt against Mark Zuckerberg? Meta employees say they don't want to train AI that may replace them

Inside Meta, a new internal move meant to push its AI ambitions forward is now facing unexpected resistance from its own workforce. The company has begun rolling out a tracking tool on work computers of US-based employees, aimed at collecting real-time data on how people use their devices. While the goal is to make AI systems smarter and more human-like in everyday tasks, the rollout has triggered unease among employees, many of whom feel they are being asked to help build tools that could eventually make their own roles redundant.

Meta employees raise privacy fears over AI tracking

According to internal communications accessed by Business Insider, the new software records keystrokes, mouse movements, click patterns, and even screen context while employees go about their daily work. This data is then used to train AI models so they can better understand how humans interact with computers, something that current systems still struggle with despite their strength in areas like coding and research.

The idea, as explained in the internal announcement, is if AI systems can observe real-world usage patterns, they can learn to replicate them more effectively. "For agents to understand how people actually complete everyday tasks using computers, we need to train our models on real examples," the post said. In practice, that means Meta employees themselves become a key part of the training loop, often without a choice. And that is where the tension begins.

The reaction internally appears far from enthusiastic. One of the most upvoted responses to the announcement reportedly read, "This makes me super uncomfortable. How do we opt out?" That question quickly became a focal point of the discussion, especially as employees realised there was no clear way to decline participation.

Andrew Bosworth, Meta’s Chief Technology Officer, responded directly in the thread, confirming that opting out is not an option on company-issued devices. His reply itself drew a wave of reactions, including shocked, crying, and angry emojis, a small but telling sign of the discomfort brewing within the company.

Meta, however, maintains that the system is being implemented responsibly. A spokesperson told Business Insider that there are safeguards in place to prevent misuse and that the collected data will not be used beyond improving AI models. The company also points out that employee activity on work devices has long been subject to monitoring, and this new tool is essentially an extension of those existing policies rather than an entirely new direction.

Still, the context matters. Meta has been aggressively investing in AI across the organisation. From setting up dedicated teams like its Superintelligence Labs to reorganising staff into AI-focused groups and hosting internal "AI Weeks," the company is clearly betting big on the technology. This new tracking initiative fits into that strategy but it also brings the human cost of that push into sharper focus.

The tool itself is not all-encompassing. It reportedly works only within a limited set of approved applications such as Gmail, GChat, Meta’s internal AI assistant Metamate, and development tools like VSCode. It also does not extend to employees’ phones. However, for many workers, the issue isn’t just about where the tracking happens, it’s about what the data ultimately enables.

There is a growing concern that by helping train AI systems to perform everyday digital tasks more efficiently, employees could be accelerating a change where fewer humans are needed for the same work. That fear is not unique to Meta, but the company's scale and influence make it a high-profile example of a bigger industry dilemma.

At its core, this situation shows a difficult balance. Companies like Meta need large amounts of real-world data to make AI more capable. But the people generating that data are also the ones most likely to feel threatened by the outcomes. And when participation isn't optional, the line between innovation and discomfort becomes even thinner.