Meta's Push to Train AI on Worker Clicks Exposes Gaps in Consent and Corporate Accountability

Tech companies have long monitored employee performance, but Meta is now taking that practice into new territory. The company has begun rolling out software across its US workforce that logs mouse movements, keystrokes, clicks and periodic screenshots from work applications. The explicit goal is to generate training data for AI models that could one day handle tasks without human involvement.

Turning Routine Work Into Training Fuel

This effort, labeled internally as the Model Capability Initiative, treats everyday employee actions as raw material for machine learning. Rather than relying solely on synthetic data or public sources, Meta wants high-fidelity examples of how people actually use interfaces: choosing options from menus, combining keyboard shortcuts, or navigating between programs. An internal note described the process as employees helping models improve simply by completing their normal duties.

The approach makes a certain technical sense. AI agents that can operate computers the way humans do could prove more adaptable than systems limited to narrow APIs. Yet it also means the company is systematically recording the digital habits of thousands of workers to speed the arrival of tools that might reduce the need for those same workers.

Privacy Risks Meet a Weak Regulatory Environment

Meta's history of privacy missteps adds weight to concerns about how this data will be handled. Although the firm says it will apply safeguards to sensitive information, it has offered few details on what those safeguards entail or how long the data will be retained. The program runs only on US employee devices, an important boundary given that similar monitoring would likely violate European Union rules on worker surveillance.

Legal experts have pointed out that American federal law currently places almost no restrictions on employers collecting this kind of granular behavioral data. That vacuum creates space for aggressive experimentation, but it also leaves employees with limited recourse. Questions linger about whether workers can decline participation without affecting their performance reviews or career prospects at the company.

Job Cuts Add a Layer of Irony

The tracking program arrives as Meta prepares to reduce its global headcount by about 10 percent in the first of several rounds of layoffs scheduled for this year. The juxtaposition is hard to ignore. Employees are being asked to generate the very data that could help AI systems replicate parts of their roles, even as the company trims its payroll.

This is not the first time a major technology firm has pursued automation while shrinking its workforce, but the explicit use of employee data to train replacement agents feels more direct. It raises the possibility that companies will increasingly view their own staff as both labor and proprietary data sources, a dual role that deserves closer ethical examination.

What the Industry Should Watch

Beyond Meta, the initiative could influence how other firms think about collecting interaction data. If the resulting AI agents demonstrate genuine capability, competitors may feel pressure to adopt similar methods. At the same time, the data's quality is not guaranteed. Mouse movements and screenshots capture behavior but not the reasoning behind decisions, the negotiation of ambiguous requirements, or the contextual knowledge that experienced employees bring to their tasks.

Regulators and policymakers have so far paid limited attention to these workplace data practices, focusing instead on consumer privacy and high-stakes AI safety. That gap matters. Without clearer standards around consent, data minimization and transparency, companies can continue expanding surveillance under the banner of innovation.

Meta is not alone in chasing more capable AI systems, but its methods spotlight an uncomfortable truth: progress in this field often depends on extracting value from people in ways they may not fully understand or control. Until those dynamics are addressed through stronger norms or rules, initiatives like this one will continue to test the boundaries of acceptable corporate practice.