These 3 Agreements Secured AI Protections for 30,000 Union Workers

Labor leaders and worker advocates have long acted to address concerns about new technology in the workplace, whether it has been the mainstreaming of automated teller machines (ATMs) in the 1990s or self-check out machines in fast food establishments or grocery stores more recently. Today, labor leaders are contending with AI developers promising new AI tools that will create entire movies, do your accounting, and assist doctors. Even with promises of improved efficiency and reduced costs not materializing as expected, these promises have driven rushed adoptions across nearly every industry. But as organizations race to capitalize on AI’s potential, the people most affected, workers, are often left out of the conversation.

The narrative around AI adoption has largely been framed as two opposing sides: employers seeking productivity gains, and workers afraid for their livelihoods. But this isn’t how the story has to unfold. Worker displacement does not have to be the cost of innovation. In fact, when workers are brought into the AI adoption process early, and their expertise is treated as an asset, the outcomes are better for workers and employers.

Labor leaders and worker advocates have a crucial role to play in how this process is shaped. Unions across industries and borders are demonstrating that it is possible to win strong AI protections for workers.

So what does it look like to do this well?

Today Partnership on AI is publishing three case studies to show how unions are navigating AI adoption. Building upon on-going work with both labor and employer leaders, these case studies help us understand how to implement the Guidelines on Shared Prosperity, which outline recommended practices for AI-using organizations for how they can incorporate worker voice as they deploy AI tools.

These three case studies feature interviews with the Financial Services Union of Ireland, SLC-CGIL (“Communication Workers Union”) of Italy, and SEIU Local 668—as well as an interview with representatives of the Commonwealth of Pennsylvania, who SEIU Local 668 negotiated with.

These three unions have won AI protections for their roughly 30,000 members across different industries and countries. Despite these different contexts, across these case studies, two overarching insights emerged:

1. Unions can build on existing labor laws and previous agreements regarding technology in the workplace, and leverage long-standing expertise, despite the ‘newness’ of AI tools. While AI may be new, the challenges it presents are echoes of existing labor challenges. Unions can look to existing labor laws, prior technology agreements, and strategic bargaining tools as a foundation for establishing AI protections, even when blocking adoption outright isn’t possible. Wins in one sector shouldn’t stay there; unions can and should use each other’s agreements as models and precedent to build on across industries.
2. Labor leaders may need additional support to design adequate long-term protections against potential job displacement. Uncertainty about the potential path of technological development has grown—for that reason, PAI recently announced new multistakeholder work which includes partnering with labor leaders to leverage economic scenario planning to help them plan for the future. The scale, speed, complexity, and uncertainty of AI’s potential development means that internal expertise alone may not be enough to protect against job displacement. Labor leaders across interviews relied upon additional technical support from allied unions or federations, or academic experts, and further technical assistance is likely needed to design adequate long-term protections against job displacement.

In addition, each case study yielded unique insights. For example, in the FSU case study, uncertainties about AI use and risks shared by employers and employees provided an opportunity for the union to frame pre-emptive protections as mutually beneficial. In the SLC-CGIL case study, we found that even if unions may not be able to “stop” the introduction of AI tools, they can make sure they are part of the governance process. The SEIU Local 668 case study highlighted the need to seek broad, high-level agreements to protect a wider range of workers, and use “patterning” or “impact bargaining” to extend protections across unions—and our interview with representatives of the Commonwealth of Pennsylvania showed how involving workers as experts to produce AI policies can improve outcomes for both employees, the organization, as well as the broader public benefiting from government services.

These case studies demonstrate what is possible when workers have a stronger voice in shaping AI adoption. The window for getting AI adoption right may be closing fast. Unions and worker-led organizations have the power to shape what comes next. The decisions being made today will determine whether AI expands opportunity or concentrates power in the hands of a few. To learn more about our work in this space sign up for our newsletter below.

We would like to thank our interviewees, as well as Massimo Mensi of UNI Global Union for his generous assistance in translating our interview and case study with SLC-CGIL between English and Italian. We would also like to thank the partners who suggested or introduced us to potential case study participants, including Christina J. Colclough (Why Not Lab), Ben Richards (UNI Global Union), Michelle Miller (Center for a Just Economy, Harvard), Daaiyah Bilal-Threats (National Education Association), and Lisa Kresge (UC Berkeley Labor Center).