Building Worker Power in an AI Economy: Sectoral Bargaining and Industry Standard Boards

Introduction

Right now, two important policy debates are near convergence but haven’t fully found each other yet in public discourse: AI’s impact on workers and labor law reform.

Artificial intelligence (AI) continues to dominate headlines and discussions across society from board rooms to state legislatures to break rooms. Americans are increasingly worried about AI’s impact on the economy and what it means for workers.¹ In response, policymakers and advocates have developed a range of plans that seem to be designed, in large part, to address job loss and transitions. But a foundational piece of any policy response to AI should be increasing workers’ power relative to their employers and creating real opportunity for them to bargain over AI. The use and impact of AI already varies from industry to industry and even from business to business. Proposals to impose new taxes on businesses’ use of AI to deter automation or fund an expanded social safety net are worth serious consideration. But these proposals alone don’t address how employers’ use of AI in the workplace impacts wages, health and safety, and job quality. Additional interventions are necessary to address the uneven impact of AI across industries and businesses. Expanding workers’ bargaining power over AI should, therefore, be an immediate and top priority for policymakers.

Preceding recent conversations about AI has been a decades-long degradation of labor law in the US.² Over the last two years, the Trump administration and Supreme Court have accelerated the erosion of workers’ rights by upending the independence of the National Labor Relations Board (NLRB) as well as attacking federal employees’ collective bargaining rights.³ In response, advocates have redoubled efforts to advance labor reform ideas that move beyond simply shoring up existing processes and rights under the National Labor Relations Act (NLRA). Sectoral bargaining and industry standard-setting have both risen to the top as promising vehicles to meaningfully improve workers’ economic outcomes.

Ultimately, workers will fare best in an economy shaped by AI if they have much stronger bargaining leverage.⁴ The reality, however, is that workers currently face some of the lowest levels of bargaining power of the last century. Our labor system is all but broken, and workers need full-scale labor law reform that builds on the proposed Protecting the Right to Organize (PRO) Act. But given the speed of the development and deployment of AI, workers would also benefit from more targeted strategies in the short term. Absent adequate interventions now, AI’s negative consequences may take further root in workplaces and our labor market, making course corrections harder down the line.

The public’s attention and the political salience of AI may create opportunities to implement AI-specific bargaining and standard-setting regimes. If implemented, these AI-standard and -bargaining models could serve as vital proof points for broader labor reform while delivering much-needed mechanisms to counter employers’ use of AI. This brief identifies and discusses two key federal and state policy interventions that would increase workers ability to influence how AI is used in their workplaces:

• Congress should lift NLRA preemption for multiemployer bargaining over AI and related technologies, clearing the way for states and localities to implement sectoral bargaining models for private-sector workers to negotiate with businesses in their industry over the use and impact of AI on their jobs and working conditions.
• States and localities should establish industry standard boards for AI, providing baseline employment protections for workers in sectors where there are common uses and impacts of AI.

Brief Background: Sectoral Bargaining and Industry Standard Boards

Decades of corporate attacks and Congress’ failure to pass meaningful reforms have all but gutted labor law in the US.⁵ Despite having clear rights to organize a union and bargain collectively, workers today face an uphill battle to realize these rights. Employers all too frequently engage in unlawful practices to deter and bust union organizing.⁶ Penalties under the NLRA are woefully weak, and adjudication processes are so slow that any relief for workers is often too little and too late.⁷ And even if workers are able to successfully organize a union, a contract is far from guaranteed, as companies can easily drag their feet or refuse to bargain in good faith with little punishment.⁸ All of this is on top of decades-long efforts by companies and anti-worker policymakers to weaken and carve out protections under the NLRA, including allowing states to pass “right-to-work” laws.⁹ As a result, union density is lower today (10 percent overall and 5.9 percent among private-sector workers as of 2025) than when the NLRA was passed in 1935 (12.8 percent).¹⁰ In other words, the law and government body that Congress created to support the free flow of commerce by “encouraging the practice and procedure of collective bargaining” has resulted in a lower share of workers today engaging in these activities.¹¹

Labor unions and advocates have long pushed for pragmatic federal labor law reform. The Employee Free Choice Act (EFCA) was first introduced by pro-labor policymakers in 2003 and, despite passage in the House in 2007 and significant efforts by advocates in 2009, didn’t secure a filibuster-proof majority in the Senate and never became law.¹² In the face of continued assaults on labor rights, advocates expanded the breadth of reform proposals with the introduction of the PRO Act in 2019.¹³ Like the EFCA, the PRO Act has passed the House, most recently in 2021, but has fallen short of earning enough support in the Senate. Over the last two years, the Trump administration’s and large companies’ heightened attacks on the NLRB have reinforced the need to advance innovative policy models and new pathways to increase worker power. Many policy experts today recognize that even if the PRO Act were passed, it alone would not ensure workers are able to fully realize the bargaining power that is needed to match the structure of today’s economy. Below is a short summary of key models that are actively being considered in labor (and employment law) spaces to complement the PRO Act and are relevant to the longer discussion of policy options for AI in the sections that follow.

Sectoral Bargaining

Perhaps the most ambitious reform idea is establishing true sectoral bargaining in the US. Sectoral bargaining is a process for a group of unions and businesses (or their associations) in a given industry to negotiate agreements related to wages, benefits, and conditions of employment.¹⁴ Any agreement reached would then apply to all employers and covered workers in the sector. For example, unions and businesses in the hospitality industry could establish industry-wide agreements for a full range of occupations (e.g., housekeepers, bartenders and cooks, laundry services, and front desk staff) that would apply to all hotels and businesses in the sector. There are numerous advantages to this bargaining model, but among the core benefits is that agreements can cover far more workers, including those who have been purposely contracted out by employers to intermediary firms or classified as independent contractors in order to exclude them from enterprise-level bargaining. Moreover, this approach largely removes the ability of firms to use low wages as a means of increasing their profits in a competitive market.¹⁵ With industry-wide wage rates and benefits, firms need to compete more on the quality of the goods and services they deliver rather than relying on a labor cost-cutting strategy to sustain profits. To the extent that firms do use wages as a strategy in competition, it is by setting their wages above industry-wide rates in order to attract and retain desired talent.¹⁶ Overall, sectoral bargaining can help reduce inter-firm wage inequality, which is the main driver of wage inequality in the US.¹⁷ Other countries have long utilized various versions of this bargaining model with much success in achieving strong agreements and supporting high union density.¹⁸

In the US the NLRA governs the process for organizing and establishing a union to represent private-sector employees covered by the NLRA at a given workplace. The NLRA does not cover public-sector workers, agricultural and domestic workers, or independent contractors. Under the NLRA, there is not a process for covered employees to easily come together and require multiple employers across a sector to bargain and establish an agreement.¹⁹ At the state level, lawmakers are currently limited in their ability to establish and oversee sectoral bargaining processes (discussed further below and in sidebar box).

Despite these significant limitations, in recent years sectoral bargaining has advanced at the state level among workers who fall outside the scope of coverage under the NLRA. Massachusetts led the way in 2024 by establishing, through a ballot measure, rights and processes for rideshare drivers to join a union and negotiate on a sector-wide basis.²² By spring of 2026, workers had successfully formed a union, secured certification by the state, and can now begin bargaining with Uber and Lyft.²³ California and Illinois have since established similar rights for rideshare drivers, and additional efforts are underway in other states.²⁴ More recently, the Center for Labor and a Just Economy (CLJE) has developed model legislation for states interested in creating sectoral bargaining structures for industries.²⁵ This model legislation addresses often-difficult implementation and design questions, including how to define sectors, the process for workers to petition the state and trigger the bargaining process, administrative funding mechanisms, options for enforcement, and how to address an impasse in bargaining, among other issues. This model could be passed and implemented today for workers in industries that, similar to rideshare drivers, are likely not covered by the NLRA. Importantly, if not for NLRA preemption, this model legislation could also be applied to a narrower scope of bargaining such as AI, as CLJE expressly notes.

Industry Standard Boards

Over the last 10 years, a handful of states have turned toward novel strategies within traditional employment law as a means to improve outcomes for workers in sectors that have particularly low wages, poor working conditions, or face unique barriers to organizing under the NLRA.²⁶ As discussed above, states are limited in their ability to pass labor laws—legislation governing the process for employers and workers to negotiate, on their own, the terms and conditions of employers. But states do have a clear authority to establish their own employment laws—minimum standards and requirements imposed on employers and rights afforded to individual workers. Industry standard boards are a specific type of regulatory process that falls within the broad category of employment law. Different variations of this model are often referred to as wage boards, standard-setting boards, or co-regulation. Here’s how it works:²⁷

States generally establish a tripartite governing board composed of representatives from government, labor and workers, and business. These boards, depending on their authority, set or recommend to state agencies a minimum wage for covered workers in an industry, and in some instances are able to engage in establishing standards for other conditions of employment, such as scheduling, safety standards, or access to benefits like paid time off. Ultimately, the state establishes and enforces a new regulatory standard based on the board’s recommendations or decisions. Unlike traditional employment standards that apply across industries within a given state, industry standard boards are able to set standards that are more tailored to the specific industry. This may mean setting a sector-wide minimum wage that is higher and more appropriate than an existing state minimum wage, or addressing unique industry practices related to, for example, scheduling or overnight shifts.

A handful of states have deployed this model for standard-setting. As of this writing, 11 states and localities have established one form of an industry standard-setting board, and these models have been used across six groups of occupations or industries, including the fast food industry, nursing homes, agriculture workers, and homecare workers.²⁸

Key Differences Between Sectoral Bargaining and Industry Standard Boards

Sectoral bargaining and standard-setting boards are often considered alongside each other as a means of reaching workers within the same sector. However, these policy models have noticeable differences and indeed are not actually interchangeable nor mutually exclusive. Just as the minimum wage is not a substitute for unions negotiating a collective bargaining agreement, industry standard boards are not a substitute for robust sectoral bargaining; the two can work together in tandem. A few key structural differences are highlighted in the table below.

Looking Toward Sectoral Bargaining and Standards for AI

With each passing week there seems to be a new narrative around AI’s impact on society and our economy. But throughout many of these conversations appears a clear desire among leaders to ensure that workers share in the gains that may be realized by AI’s productivity boom, to help workers guard against the negative consequences of AI, and to preserve dignity in their work. There are a number of proposals related to AI and workers, but to date most of them are either focused on specific negative use cases, such as AI surveillance and data collection, or more dire scenarios related to job loss.²⁹ Few proposals, however, are designed with the aim of meaningfully increasing workers’ bargaining power so they can directly influence how businesses use AI in the workplace. Sectoral bargaining and standard-setting are obvious vehicles that legislators should consider. This isn’t a new idea—others, including work by the CLJE, have pointed toward sectoral bargaining for AI for a few years.³⁰ Yet there remains a lack of serious attention being paid to labor reform by policymakers who are focused on AI, even among those who are concerned about AI’s impact on workers and have themselves previously called for exploring sectoral bargaining more broadly.

While it’s important to secure broader labor law reform, including passing and building on the PRO Act, the speed with which AI is being developed and deployed means that workers need policy solutions now. Absent the ability to pass full federal labor law reform, proponents of labor reform should consider more tailored, yet still bold, strategies that would increase workers’ bargaining power over AI.

State Sectoral Bargaining on AI

Over at least the last year and a half, Congress has considered multiple proposals to preempt state regulation of AI developers, including even the most basic requirements related to data training disclosures and watermarking of AI content.³¹ Proponents of these measures have called for preempting state regulation in the name of advancing a uniform federal regulatory framework for AI. Yet, at the same time, members of Congress have sought to create so-called sandboxes for AI where companies could receive exemptions and waivers from federal regulatory requirements in the name of further development.³²

Civil society has, so far, beaten back these efforts, protecting states’ position as the frontier of regulating AI in the US. However, those who have pushed back against limiting states’ regulatory authority have not advanced their own offensive strategies and policies to expand what is possible at the state level. Policymakers should, therefore, develop a federal bill that would lift NLRA preemption for state and local labor laws that cover multiemployer structures of bargaining over AI. At a moment when the public shares widespread concern about AI’s impact on the economy and is voicing skepticism about the US’s ability to adequately regulate AI, expanding states’ abilities to protect workers is a glaring gap in federal policymakers’ response to AI.³³

As discussed above, courts today have interpreted federal labor law as preempting states from passing their own laws that establish collective bargaining rights for employees covered by the NLRA. This blanket labor law preemption is distinct from other workplace standards. Under employment laws, workers are generally afforded federal protections, and states are able go above this federal floor. Consider, for example, the federal minimum wage, antidiscrimination protections, and even health and safety standards. To be sure, designing and implementing multi-jurisdictional labor rights comes with uniquely complicated questions.³⁴ However, pairing NLRA enterprise bargaining with state sectoral bargaining over a specific topic, such as AI, raises far fewer policy design questions.

At a federal level, the carve-out to preemption itself need not be complicated or even affirmatively set forth state-level policy designs. A federal bill would simply need to amend the NLRA to expressly note that states and municipalities arepermitted to create rights and processes related to multiemployer bargaining models on the subject of AI and related technologies, including algorithmic management tools.³⁵ As discussed further below, this approach would preserve employees’ rights under the NLRA while affording states the ability to create new collective bargaining models that would ultimately undergird enterprise bargaining. Specifically, states would be able to establish sectoral bargaining laws, such as those proposed in model legislation developed by the CLJE, that enable unions to bargain over the use and broader impacts of AI. A range of AI and related technological topics could be bargained over, which might include: corporate governance requirements; notice and rights related to data collection; limits and rules on automated decision-making, including hiring, performance reviews, wage setting, and terminations; specific standards related to common industry-specific uses of AI such as setting pace-of-work requirements or recommendations of patient care; and severance for AI-related job displacement, among other topics.

The concept of lifting NLRA preemption (as well as sectoral barging proposals) is at times a flashpoint in labor law reform discourse. This debate, however, has often been shorthanded to a point where it limits real consideration of how preemption might be lifted and the impact of specific policy design choices (e.g., “Do you favor lifting preemption or preserving it?”). A few important, but nonexhaustive, points on preemption and sectoral bargaining should be noted.

Before NLRA preemption doctrine—and even since—states have designed and implemented collective bargaining statutes. Lifting NLRA preemption, even partially, may feel to some like wading into unchartered waters. However, Gali Racabi has recently assembled a thorough accounting of state labor laws that were passed but are now dormant because of preemption.³⁶ Specifically, in the wake of the NLRA being passed and upheld by the Supreme Court, 18 states passed their own laws affording workers with collective bargaining rights. In the years since the Supreme Court’s Garmon decision, states have still stood up and administered bodies that regulate labor rights for workers excluded from the NLRA, most commonly public-sector unions and agricultural workers, and more recently rideshare drivers, as discussed above. The point here is that states are not without capacity or familiarity in designing and implementing laws that afford workers collective bargaining rights. These muscles may have been weakened under NLRA preemption, but recent advancements in the rideshare industry indicate an eagerness to rebuild strength.

Lifting NLRA preemption is not all or nothing. NLRA preemption could be narrowed or eliminated primarily through two mechanisms (though other more theoretical or technical channels exist, too).³⁷ First, the Supreme Court could overturn Garmon and Machinists preemption (or effectively do so if parts of the NLRB or NLRA are found unconstitutional). Second, Congress could pass legislation that explicitly allows states to regulate topics and actions currently governed by the NLRA. Under either approach, there are a variety of ways to define the scope of newly permissible state labor laws and how they would interact with federal labor law. One could imagine, for example, a scenario in which Congress broadly removes existing NLRA preemption in a way that affirmatively allows states to establish rights in parallel and beyond those provided to employees and employers under the NLRA, as well as regulate specific actions by parties that are not fully addressed under NLRB rules. Indeed this is, in some ways, how the states’ “right-to-work” laws function now. Under the NLRA, unions and employers are allowed to include in collective bargaining agreements a requirement for employees in a bargaining unit to pay union dues shortly after being hired. However, in 1947 under the Taft-Hartley Act, Congress explicitly allowed states to pass their own laws that prohibited employers and unions from reaching such agreements, effectively supplanting what is permissible under the NLRA.³⁸

Among those who are hesitant to lift current NLRA preemption is a concern that while it creates significant opportunities for states to expand and support workers’ abilities to engage in collective bargaining, it may also lead some states to further empower employers. As Ben Sachs and John Fry have explained, in the absence of preemption pro-worker states could, for example, create stronger penalties for employers’ violations of workers’ rights, or require interest arbitration to reach first contracts. Other states, however, might attempt to allow employers to fire workers on strike or decline to bargain with unions altogether.³⁹ Importantly, Sachs and Fry believe that even when weighing these considerations, lifting preemption would, on net, benefit workers.⁴⁰ Opponents of lifting preemption often focus exclusively on the potential for states to weaken labor rights relative to the NLRA, viewing it as a red line that shouldn’t be crossed. It seems they take an “all workers better off or no workers better off” approach to reform—a noble and understandable approach, particularly in an area of law that has historically excluded and, by design, left behind entire segments of the workforce that were predominantly workers of color.⁴¹ Lifting preemption, though, does not necessarily mean creating an opening for some states to reduce workers’ rights. Like employment standards, the NLRA or any future federal labor law could serve as a floor that states could not go below but could build upon. While Racabi points out that there are implementation challenges in the labor context where the NLRA affords both workers and employers rights, it is not impossible. Indeed, Racabi, Cynthia Estlund, Andrew Elmore, and the CLJE have all conceived of a few approaches, including, in one instance, a model akin to OSHA’s review and certification of state health and safety laws.⁴²

The challenges associated with establishing both federal and state labor laws while maintaining a federal floor of rights arises most acutely where the models of bargaining directly overlap. Most policy debates assume lifting preemption in a manner that affords states and cities the ability to directly regulate enterprise-level bargaining. As described in this proposal, this need not be the case. Congress should, as a first step, lift preemption to enable states to pass laws establishing multiemployer bargaining structures that complement enterprise bargaining without wading directly into the center of the NLRB’s jurisdiction, as Sharon Block and Seema Nanda have previously proposed.⁴³ In this way, states could strengthen workers’ rights and bargaining power by establishing processes to reach industry-wide contracts; enterprise-level bargaining would then stack on top of these agreements and continue to be governed by the NLRB.

Sectoral bargaining can complement, not supplant, enterprise bargaining. A key consideration if preemption does occur is how the creation of sectoral bargaining might impact unions’ efforts to increase membership and expand enterprise-level bargaining. By design, sectoral bargaining covers a much broader set of workers in a sector than those who are union members. Some advocates query whether establishing sectoral bargaining would indeed help unions expand enterprise bargaining and union membership if sectoral bargaining delivers many of the same benefits. In general, there is likely a complementary connection between these two models, as success at one level can fuel momentum at the other.⁴⁴ In the context of this AI-specific proposal, there’s likely an even stronger connection given that the scope of bargaining is clearly very different. Take, for example, unions that have successfully negotiated employer notice requirements when using technologies to monitor employees’ work activities, or contract terms pertaining to how automated systems are used for timekeeping and compensation.⁴⁵ These unions are likely eager to expand these baseline standards to other workplaces; a sectoral model would be a clear vehicle to do so and in a more expedited way than organizing and bargaining worksite-by-worksite.

Sectoral bargaining would also provide unions with a new point of connection to nonunion workers. While sectoral bargaining doesn’t require new organizing and membership, some experts have pointed out that the success of multiemployer bargaining today is strongest in sectors with higher union density.⁴⁶ It follows then that unions negotiating a sector-wide agreement with businesses will have more leverage at that table if there is either already strong union membership in the industry or broader engagement from workers.⁴⁷ On a topic like AI where there is already a high level of public concern, it is easy to imagine engaging and mobilizing workers—both union and nonunion—during industry-level bargaining. Through this process unions may be able to translate that engagement by nonunion workers to support nascent enterprise organizing campaigns.⁴⁸ This may be a particularly beneficial tactic within industries, like the service sector, where business structures and practices have created unique challenges to organizing. In short, as unions are aiming to organize more workers in new workplaces, creating additional avenues to secure benefits for workers on one of today’s most pressing issues should be a welcome boost to enterprise organizing and bargaining efforts.

While some labor experts and advocates may view such a carveout of preemption as an unprecedented move, the reality is that there are already a number of preemption exceptions, including the “right-to-work” laws. Lifting preemption for states and cities to advance sectoral models of bargaining on AI would, in practice, be a way for federal policymakers to begin seriously considering how to advance state labor laws in ways that are most beneficial to workers, while making meaningful progress in the short term on AI-related issues. At a moment when both support for unions is at an all-time high and workers are expressing concern over AI, policymakers and advocates should be eager to catch this lightning in a bottle. In doing so, workers who benefit from these AI-specific bargaining models may be more interested to join unions and subsequently build the power necessary to push for the broader labor reform that is so desperately needed.

Industry Standard Boards

The development of AI continues at a seemingly breakneck pace while the adoption of AI and other technologies by businesses varies considerably across sectors. According to Census Bureau data from early 2026, about 20 percent of all businesses have adopted AI in some fashion. Within certain industries, such as financial and professional services, adoption is higher at 30 percent, and in other sectors, like retail trade, it still remains surprisingly low at 14 percent.⁴⁹ Though it is important to recognize that a larger share of workers in these industries likely use or are subject to AI given these data reflect the percentage of businesses—not workers—that use AI. Across these and other industries there are already common use cases of AI and trends of its impact across enterprises. In the warehousing sector, advocates have long been calling for policy interventions to address the use of algorithmic technologies that set worker productivity targets, which have led to a spike in workplace injuries.⁵⁰ In the transportation sector, use of AI and related technologies is so commonplace that unions representing more than 16.5 million workers worldwide recently created a global framework to address their concerns, including excessive and blackbox surveillance practices.⁵¹ These and many other patterns of industry-specific use cases are a strong reason for states to utilize sectoral standard-setting boards when regulating AI in the workplace. In establishing these boards, states should keep a few key policy design considerations in mind, as others have detailed in earlier work.⁵²

Today, states and cities have created sectoral standard-setting in only a handful of industries.⁵⁴ In what industries then should policymakers first advance this regulatory process for AI? State legislators would likely need to first advance a standard-setting board in one or two sectors in their state before being able to deploy the strategy more broadly. To that end, policymakers should consider the following guiding principles when deciding which sectors are ideal as first movers and successful proof points for this form of regulatory process.

First, they should consider industries that are early and high adopters of AI and where clear policy intervention needs have already surfaced. This would make it easier to define the scope of regulatory authority and the specific topics to address through standard-setting. A few industries serve as clear examples of where the adoption, trends, and worker concerns have all reached a level that is ripe for sectoral standards, including warehousing and ecommerce, telecommunications, healthcare, and financial services, among others. Second, they should consider industries where public concern for businesses’ use of AI is already high and would likely yield support for new standard-setting boards; the coalition demanding intervention in these instances could be broader than simply the workers directly affected. In many instances the public already recognizes that workers are the front line of defense for how AI’s use in businesses ripples beyond companies’ walls and impacts society. For example, how a hospital deploys AI and allows practitioners to override decisions they know are not in the best interest of their patient is of significant concern to the public. Similarly, how employees within the financial industry use AI or other automated systems to approve or deny loan applications is of significant interest to consumers. And in telecommunications, the ability to eventually speak with a live agent is a perennial headache for customers. Proponents of greater worker protections should, therefore, leverage the public’s concern with AI’s use and impact to garner broader support for the establishment of sectoral standard boards. Finally, policymakers should consider to what degree policy experts, unions, and workers have already identified and piloted specific solutions to problems that would likely be raised in the sectoral board setting. Setting up boards for quick success will likely be vital to subsequent boards being established. To that end, to the extent that policymakers are limited in their political capital they should move to establish AI sectoral standard-setting boards, where there is a clear pathway to timely consideration and adoption of new standards.

As with preemption (and sectoral bargaining), some policy experts have reservations about putting limited resources behind the establishment of new sectoral standard-setting rather than toward jurisdiction-wide rights or expanded labor rights. Why, some might ask, should states create a structure for raising the minimum wage only for fast-food workers, when low-wage retail workers also need a raise? Here, too, debates have been conflated to the point where it’s easy to lose sight of larger aims for improving workers’ outcomes and how sectoral standards—in conjunction with other tactics—help achieve that objective. At the risk of shortchanging these dynamics, below are a few key points that are worth calling out.

Sectoral standard-setting is not a substitute for collective bargaining. At the most basic level, employment and labor law need to work in tandem to maximize worker power. The purpose of labor law is to create a process for workers to aggregate their individual power. As discussed above, the exclusive pathway to collective bargaining for most private-sector employees has been gutted, severely limiting the ability of workers to aggregate their power when it is needed most. Meanwhile, individual worker power has also been significantly reduced by companies through a range of business practices that push workers outside the scope of employment protections and further away from businesses that actually control workers’ economic outcomes. Therefore, both employment and labor reforms need to be pursued in tandem in order to build stronger collective bargaining power. Fixing one alone won’t deliver the full potential of aggregate worker power. Sectoral standard boards that exist alongside collective bargaining would allow states to establish baseline protections across an industry; this in turn enables enterprise bargaining to focus on leveraging collective market power to go above and beyond these basic standards and achieve greater economic rebalancing from businesses. Consider, for example, how unions have played a crucial role in helping to first establish industry-specific health and safety standards at the state level and then set their sights in bargaining on bigger gains. In California, nurses’ unions were vital to establishing workplace violence and prevention standards, and have since been able to bargain for even greater protections with individual hospitals and focus their bargaining efforts on key economic terms in their agreements.⁵⁵ In the context of AI and in the absence of baseline protections for all workers, unions are forced to expend their limited bargaining power to secure even the most basic protections, such as prohibiting the use of surveillance technologies in areas that would violate the privacy of workers, such as restrooms, locker rooms, or break rooms. If state standard boards establish these and other standards, then unions could focus their bargaining on bigger wins, such as requiring employers to bargain mid-contract over the introduction of new technologies. Additionally, if standard boards adopt new disclosure requirements related to the use of AI technologies, this information sharing can be used to support unions during enterprise-level negotiations.

Sectoral standards can, in some instances, be a more effective tool than traditional regulation. At a time when regulation of AI use is generally still nascent and there is widespread attention on the issue, some may wonder why new standards related exclusively to AI should be pursued through tripartite boards rather than through traditional regulations. In some states, like California, state legislators have successfully passed a range of protections that cut across industries. Where there are possibilities to pass strong jurisdiction-wide standards for workers, policymakers should pursue them. However, industry standard boards hold an advantage when one-size requirement won’t fit all industries, or when moving industry-specific requirements through the traditional regulatory process would ultimately weaken the standard.

By constituting a board to establish standards for a specific industry, state lawmakers are able to bifurcate debates about whether workers need greater protections from the specific regulatory design that is ultimately created. At a moment when deep corporate pockets are funding opposition to AI regulation, moving industry-specific safeguards through statehouses will likely face headwinds that result in a narrower or watered-down version of what is initially introduced. Moving new worker protections through a standard-setting board would keep statehouse debates more focused on the need for AI-related regulations, without yet getting into specifics. This strategy is more than just theory. Others have written about how the standard-setting boards that exist today have been successful in delivering higher wages and protection for workers, in part because they are able to more easily overcome opposition tactics used in traditional legislative debates.^[28] In short, the advantage of industry boards lies in the potential breadth of industry-specific standards and the depth of requirements that are developed through the board processes.

Additionally, the workplace impacts of AI can feel at times like they fall into different areas of traditional regulation (e.g., antidiscrimination, health and safety, wage and hour, or even occupational licensing). An added advantage of industry boards would be to empower a single set of actors to consider these issues holistically and develop a suite of standards for an entire industry.

Sectoral standard boards can bolster federal advocacy. The establishment of state sectoral standard boards could help support federal advocacy efforts. Specifically, as advocates aim to advance a range of federal policy interventions, state boards would, at minimum, provide specific policies to draw from. More importantly, their structures and manner of engaging workers could also be tapped into to support federal efforts as federal proposals gain momentum. For example, an industry standard board’s public hearing requirements would create a set of public testimony that other policymakers could more easily draw from to inform their own policy design. Moreover, the standard-setting process at the state level would, ideally, also support worker organizing more continuously than traditional state regulatory processes. Unlike a campaign to pass a specific bill in which workers may mobilize to support its passage but then dissipate, a sectoral standard board’s ongoing regulatory process and public engagement structure may help workers sustain ongoing organizing efforts, which could then be more easily mobilized toward federal efforts when they arise.

Conclusion

Workers in the US are clearly and loudly expressing concern over businesses’ development and deployment of AI. While some policymakers have taken these concerns seriously and have designed bold proposals to slow the development of AI, limit its negative impacts on our communities and environment, and design novel taxes to expand the social safety net, a glaring gap remains in federal policymakers’ response: empowering workers to directly and meaningfully influence their employer’s use of AI and its impacts on their jobs.

The sad reality is that workers today have to fight tooth and nail to realize basic rights to organize and join a union. Labor law reform is long overdue in the US. But even leading proposals today, like the PRO Act, would likely fall short of providing workers the full rights and protections needed to counterbalance the power of employers, particularly in the context of AI. Even bolder labor reform ideas are also unlikely in the short term. As AI’s impact on our labor market continues to take root, workers need viable pathways to improve standards and bargain with employers over their use of AI. Congress should, therefore, lift NLRA preemption for multiemployer bargaining over AI so that states can advance new sectoral models of bargaining. Additionally, states should consider creating industry standard boards to identify and implement new regulations related to AI.