Labor Organizing and AI Surveillance in the Workplace

January 14, 2024 by Grace Scott

In late 2022, Jennifer Abruzzo, general counsel of the National Labor Relations Board (NLRB), wrote a memo warning that artificial intelligence-enabled monitoring of labor organizing activities might violate the rights granted to workers by Section 7 of the National Labor Relations Act (NLRA).[1] While electronic surveillance pervades American workplaces and artificial intelligence increasingly performs tasks formerly given to people throughout all sectors (both serious issues in their own right), AI is particularly threatening in the labor organizing context given employers’ propensity for retaliation and intimidation in dismantling workplace unionization campaigns.[2] However, given the NLRA’s protections of union activity are uniquely potent among rights in the workplace, AI surveillance of labor organizing may be particularly vulnerable to regulation. The NLRA guarantees that workers be free from hindrances to labor organizing; any form of threat or retaliatory adverse action by an employer qualifies as a prohibited unfair labor practice under Section 8.[3] Despite the NLRB general counsel’s warning that AI surveillance undermining Section 7 rights is necessarily unlawful, there is evidence that employers have already been using AI to undermine unionization in the workplace.[4] As AI continues to grow in importance and availability, enforcement of the NLRA against such employers, as urged by the general counsel of the NLRB, will be crucial for the future of the labor movement.[5]

The fall of 2023 marked several public victories for the labor movement: The SAG, WGA, and UAW strikes and subsequent contract renegotiations put unions in the headlines more than any time in recent memory.[6] In an era of growing corporate wealth and power, workers have seized labor organizing as a tool of resistance to the pressures their employers have placed upon them.[7] At the same time, however, employers have pushed back, using surveillance tools (including AI technology initially designed for military use) to undermine workplace unionization drives.[8] Employers have adopted AI to determine employees’ propensity to organize, to monitor communications among employees, and to administer quotas to systematically disadvantage union supporters (even where the quotas are facially neutral).[9]

This sort of anti-union use of AI monitoring is an open secret at workplaces like Amazon, which are already defined by both traditional and AI-enabled surveillance.[10] In workplaces where every step, every conversation, and every bathroom break is recorded and retained, what’s to stop employers from merely adjusting how they use that information to detect union organizing? Indeed, monitoring workers is no longer the realm of especially controlling employers but is now the norm: It seems a majority of employers currently employ some sort of tracking technology, a shift which appears to have been inspired by the changing workplace norms brought on by the pandemic.[11] Whether or not safety was their true motive, employers took the opportunity presented by the pandemic to introduce new and more invasive forms of surveillance: Amazon, for example, introduced a system of AI-enabled cameras to track employees’ whereabouts, purportedly to ensure social distancing in its plants.[12] Amazon, which uses extensive camera tracking (both AI and human-supervised) to monitor its warehouse employees, also counts time spent interacting with other workers as “time off task,” a metric that, if logged for more than 30 minutes in a single day, can result in a written warning.[13] Even when surveillance is not explicitly directed at union organizing activity, it targets the same activities necessary to union organizing: Interaction with other employees (often at the job site) is necessary to collecting the signatures needed to begin seeking union representation.[14] Additionally, remote workers’ interactions with other employees may be equally subject to AI scrutiny for organizing tendencies: Interactions between some workers have been tracked using AI initially designed to map terrorist cells, put to new use to detect groups of workers interested in seeking union representation.[15]

Under the NLRA, employer practices that burden Section 7 rights must be weighed against employers’ legitimate interests; the latter must outweigh the former in order to continue.[16] Troublingly, however, those selling AI surveillance systems designed for monitoring factories tout them as tools for ensuring compliance with extant employment law.[17] Vantiq and SmartCow, two companies selling ready-to-use AI cameras for factory monitoring, claim in their advertising materials that AI monitoring will detect injured workers but fail to mention the rest of the (NLRA-protected) movements the systems will inevitably flag.[18] Even the European Agency for Safety and Health at Work (EU-OSHA) claims that AI monitoring of the workplace will help to improve worker safety through PPE compliance, among other mechanisms.[19] The safety justification is problematic if used as a pretext for more pervasive monitoring that impinges upon employees’ rights to organize. The NLRB weighs legitimate justifications employers might have against the burden on labor organizing activities, meaning that even if employers assert the necessity of such systems, their impact on workers’ rights under the NLRA might limit how AI can be used in the workplace.[20]

The Department of Labor (DOL) has a new opportunity to make exceptionally clear through rulemaking that employers’ burdening of protected labor activity through AI surveillance violates the NLRA. The Biden administration issued an executive order on the future of artificial intelligence on October 30, 2023, Section 6 of which gave DOL 180 days to promulgate a rule clarifying the protected-activity implications of AI surveillance and reaffirming existing law in the face of AI.[21] DOL and the NLRB now both have the chance to push back against the trend of workplace AI surveillance and its detrimental effect on labor organizing: With the future of work itself so uncertain given the rising dominance of AI, labor will be more critical than ever.

[1] Electronic Monitoring and Algorithmic Management of Employees Interfering with the Exercise of Section 7 Rights, NLRB Memorandum GC 23-02, at 1 (Oct. 31, 2022), available at

[2] Id. at 1.

[3] Id. at 6.

[4] Gabriel Grill & Christian Sandvig, Military AI’s Next Frontier: Your Work Computer, Wired (Jun. 22, 2023),

[5] Id. at 1.

[6] Paul Prescod, US Labor Is Having a Movement Moment, Jacobin (Nov. 21, 2023),

[7] Id.

[8] Grill & Sandvig, supra note 4.  

[9] Electronic Monitoring and Algorithmic Management of Employees, supra note 1, at 5.

[10] Lauren Kaori Gurley, Internal Documents Show Amazon’s Dystopian System for Tracking Workers Every Minute  of Their Shifts, Vice (Jun. 2, 2022),

[11] Wendi S. Lazar & Cody Yorke, Watched While working: Use of Monitoring and AI in the Workplace Increases, Reuters (Apr. 25, 2023),

[12] Tom Simonite, Amazon Touts AI for Social Distancing Amid Worker Complaints, Wired (Jun. 18. 2020),

[13] Gurley, supra note 10.

[14] Electronic Monitoring and Algorithmic Management of Employees, supra note 1, at 9.

[15] Grill & Sandvig, supra note 4.  

[16] Electronic Monitoring and Algorithmic Management of Employees, supra note 1, at 7.

[17] Tomas Chamorro-Premuzic, Can Surveillance AI Make the Workplace Safe?, MIT Sloan Mgmt. Rev. (Aug. 4, 2020),

[18] See Vantiq’s claims about its factory safety AI technology at and SmartCow’s at

[19] Nick Warburton, EU-OSHA Publishes Three Policy Briefs on Smart Digital Systems for OSH, IOSH Magazine (Jun. 5, 2023),

[20] Electronic Monitoring and Algorithmic Management of Employees, supra note 1, at 8.

[21] Exec. Order No. 14110, 88 Fed. Reg. 75191, at 75210 (Oct. 30, 2023).