In the 2025 book “If Anyone Builds It, Everyone Dies,” the authors confidently predict how superhuman AI would lead to the annihilation of humanity. The book is not fiction. Existential threat or not, AI is here and growing all around us including, increasingly, in the workplace. Its prevalence raises significant issues for unionized employers, such as whether and how AI may trigger mandatory bargaining regarding its use.
Not surprisingly, legislatures are also beginning to assess AI’s current and future impact on employees and the workplace, union and non-union. While legislation is currently under consideration in Minnesota it seems unlikely to reach the necessary bi-partisan support to pass this session. Employers would nevertheless be wise to monitor the bills and watch for new AI obligations and restrictions in the not-so-distant future.
Employers in the unionized setting are already seeing an increase in proposals from unions looking to reign in AI and to eliminate, or at least limit, its impact on workers. Recent examples of union successes include a new provision in an International Longshoreman’s Association contract that flat out prohibits all fully automated technology, and a Las Vegas Culinary union agreement that contractually obligates covered employers to bargain over any decision to implement AI in the workplace.
Most unionized employers understand that matters affecting the “terms and conditions” of employment such as wages, hours, benefits, etc. are mandatory subjects of bargaining. So, if an employer decides to reduce wages or cut employee hours, whether connected to AI or not, those must be negotiated. But what if, as widely predicted, the introduction of AI reduces the need for human labor, leading an employer to implement layoffs? The NLRB has not yet issued a definitive ruling on bargaining obligations from AI-induced layoffs. Such employer decisions would certainly require the employer to bargain over the impact the layoffs have on employees. Unions have very little leverage in such “effects” bargaining.
Whether the decision itself would be subject to bargaining might ultimately depend on the employer’s motivations behind its adoption of AI and the associated layoffs. Is it primarily to reduce labor costs? If so, the decision itself is likely subject to bargaining. Even if not, unions might successfully argue bargaining is required because such actions are akin to subcontracting. It also may be the case that bargaining is determined to be mandatory under various provisions of an existing collective bargaining agreement.
Regardless, the NLRB and the courts are sure to provide answers in the coming months and years.
Meanwhile, the Minnesota legislature is presently considering legislation that would, if passed, impose new limits on all employer use of AI. Senate File 4689 seeks to regulate the use of what it calls “Automated Decision Systems” (ADS). It would essentially cover all employment-related decisions relating to the implementation of AI. It would require advance notice of, and employee consent to, the use of ADS, would impose significant recordkeeping obligations, and employees would have the right to know when and how ADS influenced “adverse” employment decisions. Employees would have appeal rights to challenge those decisions. And of course, the legislation would provide employees with the right to sue employers over violations, obtain damages (including punitive damages), attorneys’ fees, and penalties of up to $2,500 per violation per day.
As we noted above, the current makeup of the Minnesota legislature means that a bill like this is unlikely to pass this session. However, as AI’s impact on the workplace accelerates exponentially, it is all but guaranteed that laws and regulations are coming.
Unless artificial superintelligence kills us all first.