AI surveillance sparks global worker backlash, lawmakers respond

The late-2026 backlash did not arise against a novel practice; it was a response to surveillance that had already become embedded in management. By 2019, 50% of 239 large corporations employed nontraditional digital surveillance methods such as call recording, email and social-media analysis, and meeting participation tracking, up from 30% in 2015, and after the COVID-19 shift to remote work in March 2020, companies replaced in-office supervision with software tracking keystrokes, screenshots, application use, and login times to maintain oversight, creating the preexisting trend that lawmakers later moved to constrain.

By September and October 2026, that response had become visibly multi-jurisdictional. California’s Appropriations Committees held suspense-file hearings on August 13, 2026, where roughly 30 remaining AI bills either advance or die for the session and bills must reach Governor Newsom’s desk before the September deadline; California had already set the pace when it signed the Transparency in Frontier Artificial Intelligence Act, SB 53, into law on September 29, 2025, while Spain was forcing employers to tell staff about workplace algorithms, showing governments were converging on transparency and oversight requirements.

AI workplace surveillance is no longer marginal. People Matters reports monitoring has moved from an emergency management tactic to a mainstream workplace practice, and Gartner research cited via Computerworld found 74% of employers used online monitoring tools and 61% used AI-driven performance evaluation systems. A 2023 study by Janey Zheng and colleagues found that employees who felt managers were closely watching their activities reported lower levels of trust from supervisors, and those perceptions were associated with lower energy levels and greater emotional exhaustion.

The spread is also opaque and disciplinary. Coworker and Data & Society’s Labor Futures Initiative warned that workers often lacked visibility into what information was being collected and how it was being used, while a Minderoo-commissioned poll found 43% of Australians were aware some employers use technology including AI to monitor workers, but only 9% were aware monitoring was used in their own workplace. That opacity is feeding disputes, from a cashier given a final written warning because the AI monitoring system had flagged her making too many errors, with the Shop, Distributive and Allied Employees Association saying the reported error rate was 0.18%, to Eurofound’s finding that nearly 20 % of EU employees report being monitored by a computer programme, 18 % report being allocated their tasks by a computer and 10 % say that it determines their working hours.

The harm workers describe is not just being watched, but being governed by systems they cannot see or contest. TechRadar captured the ambient fear in a stark headline — “Half of workers worry AI will still take their job — as new report claims use of agents soars 90% in just a year” — while Social Europe reported that in call centres, a November 2025 survey of more than 700 workers found 81.3 per cent of agents are under constant voice recording, typically with sentiment analysis, leaving staff facing unexplained score drops that can hit variable pay. Similar patterns appear beyond offices: a joint study by the ifo Institute, IAB, and the University of Konstanz found factory workers “take roughly 20 percent more sick days once networked machinery and robotics arrive on site,” identifying stressors linked to digitally managed work, while a KU Leuven survey of more than 2,800 employees in Belgium's metal and textile sectors found about 70 percent had little or no say in how the systems were introduced.

That opacity is stressful precisely because the rules are embedded in tools and workflows workers cannot parse: one explainer described a system that “runs whisper,” “transcribes every… all the audio in real time,” “publishes all the transcripts to mqtt,” and lets an agent collect the last five, 10 or 15 minutes for summaries and judgments. In the KU Leuven survey, only 25 percent knew how management was using the systems, underscoring how little visibility workers often have into the logic shaping their jobs. Yahoo, citing Daniel Ravid’s 2022 meta-analysis, said there was “no correlation between levels of monitoring and employee performance,” while a 2020 study found monitored truck drivers were more likely to crash — evidence that surveillance can intensify strain without giving workers meaningful clarity or control.