Meta staff protest mandatory mouse-tracking software used to train AI

TL;DR:

• Meta employees are circulating internal flyers and a petition protesting mandatory software that tracks their mouse movements and keystrokes to train Meta’s AI tools.
• The software, called the Model Capability Initiative, was installed on staff computers last month and coincides with Meta cutting 10% of its workforce “to offset” record AI infrastructure spending.
• The tracking is currently US-only, but UK Meta staff are calling for labour organisation under United Tech and Allied Workers (UTAW), a branch of the Communication Workers Union.

A protest is taking shape inside Meta over software the company has installed to capture how employees actually work — and use that data to train the AI systems being positioned as their replacements. Internal flyers distributed in toilet stalls, meeting rooms and atop vending machines ask staff: “Don’t want to work at the Employee Data Extraction Factory?” The petition, first reported by Reuters and Wired, calls out the practice of “nonconsensually extracting their data for the purposes of AI training.”

The software, and the timing

The tracking software, called the Model Capability Initiative, was installed on staff computers last month. When certain applications are in use, it captures “real examples” of how people use their computers — “mouse movements, clicking buttons, and navigating dropdown menus” — feeding the resulting data into Meta’s AI training pipelines. The premise is that high-fidelity capture of expert human computer behaviour is a useful training signal for autonomous agents.

The timing is what has galvanised internal opposition. The mandatory installation coincided with Meta announcing it would cut 10% of its workforce, a move the company said was in part to “offset” its record AI infrastructure spending. The collision of those two facts — forced AI training on the work of people whose jobs are being cut to fund the AI build-out — is the petition’s emotional core. One Meta engineer, in a post seen by nearly 20,000 colleagues this week, wrote: “Selfishly, I don’t want my screen scraped because it feels like an invasion of my privacy. But zooming out, I don’t want to live in a world where humans — employees or otherwise — are exploited for their training data.”

The UK organising response

US employers have broad legal discretion to monitor employees, but have historically relied on volunteers or paid recruits to collect AI training data — making mandatory tracking a notable escalation. The tracking is US-only for now. But Meta staff in the UK are calling for labour organisation under United Tech and Allied Workers (UTAW), a branch of the Communication Workers Union, which has confirmed the organising drive.

UTAW organiser Eleanor Payne framed the UK push as a response to a triple-stress combination: “devastating job cuts, draconian surveillance, and the cruel reality of being forced to train the inefficient systems being positioned to replace them.” For UK readers, the unionisation angle is the operative story: collective bargaining is one of the few mechanisms with the legal weight to put boundaries on AI training data sourced from employee behaviour.

Looking forward

The Meta protest is an early instance of what is likely to become a broader pattern. As frontier AI labs and large enterprises pursue agentic AI that can drive computer interfaces, the most valuable training signal is real human computer use — and the most direct way to obtain it at scale is from current employees. Whether UK employment law, GDPR provisions on employee monitoring, or sectoral collective agreements provide adequate guardrails will be tested first by cases like UTAW’s organising drive. The bigger question for UK SMEs is whether their own internal AI deployments will face similar scrutiny — and whether they will have clearer answers than Meta has so far provided.