US Public Sector Unions Secure Guardrails as 70% of Agencies Adopt Generative AI
Labor unions across the United States are successfully negotiating for human oversight and retraining programs as nearly 70% of government agencies integrate generative AI into their operations.
The Surge of AI in Government
Generative AI has rapidly become a staple in the United States public sector. According to a recent survey by Ernst and Young, nearly 70 percent of US government agencies are now utilizing generative AI tools. This widespread adoption has prompted a significant response from labor unions, which are moving to establish guardrails against potential job displacement and intrusive surveillance.
As employers increasingly look toward automation to reduce costs, labor researchers—including Kung Feng at UC Berkeley—note that the pressure on the workforce is intensifying. In response, public-sector unions are securing a series of legislative and contractual victories to ensure that technology serves to assist rather than replace human workers.
Strategic Wins for Labor
Labor organizations across several states have successfully negotiated protections that prioritize worker stability. In Oakland, the IFPTE Local 21 secured a commitment to retraining programs instead of AI-related layoffs following a strike vote in October.
In Pennsylvania, the SEIU Local 668 reached an agreement with Governor Josh Shapiro to guarantee human oversight during a pilot program involving OpenAI’s ChatGPT Enterprise. Similarly, the Transport Workers Union (TWU) in Ohio has gained the power to veto the implementation of autonomous vehicles, ensuring that safety and human expertise remain at the forefront of transit operations.
Legislative Protections and "Bossware" Bans
State governments are also beginning to codify these protections into law. In September, California Governor Gavin Newsom enacted a ban on AI surveillance software, commonly referred to as "bossware," and prohibited the use of automated systems for firing employees.
New York has taken similar steps, with laws in place since late 2025 that protect state employees from being displaced by AI. These legislative moves represent a growing trend of state-level intervention to prevent the unchecked use of automation in the workplace.
The Demand for Mandatory Human Control
The overarching strategy for major labor leaders, including Liz Shuler of the AFL-CIO, Lorena Gonzalez, and TWU’s John Samuelsen, is not to reject AI entirely but to demand "mandatory human control." While these leaders support AI safety systems that can aid workers in their daily tasks, they insist that every AI application in the workplace must remain under human supervision.
As generative AI continues to permeate the public sector, the focus of US unions remains clear: securing a future where technology is a tool for enhancement, governed by strict human-in-the-loop requirements.
