The SAG-AFTRA strike of 2023 was, in retrospect, a signal event — not primarily because of its outcome (the contracts were mixed, the studios retained significant AI usage rights while conceding others) but because it was the first large-scale collective bargaining action explicitly centered on AI labor rights. The entertainment industry unions saw the threat clearly and mobilized quickly. What happened afterward — in those industries, and in the broader organized labor context — tells you something real about the relationship between AI and union power in 2026.
The complicating truth is that AI is simultaneously the biggest structural threat to union labor in several decades and, in certain ways, creating conditions that are more favorable to organizing than they’ve been since the 1980s.
Start with the threat. Unions derive their power from the credible threat of withdrawing labor, which requires that labor be difficult to replace. AI directly attacks this foundation. In any sector where AI can credibly perform a significant share of union work — content production, data processing, logistics coordination, basic customer service escalation — the threat of substitution weakens labor’s bargaining position. Even if the substitution isn’t economically optimal right now, the credible future possibility of substitution changes today’s negotiating dynamics. The employer who could replace 40% of your workforce with AI tools next year doesn’t have to do it to extract concessions; the threat alone shifts the bargaining equilibrium.
This is why the 2025-2026 contract cycles in media, finance, and professional services have been difficult for unions. The Writers Guild of America West was back at the table in early 2026 on streaming residuals, and the studios’ negotiating posture was notably harder than it had been in 2023 — partly because the 2023 agreements gave them time to invest in AI workflows that have genuinely reduced their dependency on certain categories of writing labor, and partly because the macroeconomic environment made them less eager to settle quickly. The leverage dynamic shifted.
Now the complicating part.
AI deployment creates legitimate, specific grievances at scale, and legitimate specific grievances at scale are the raw material from which organizing campaigns are built. The healthcare workers who are being required to supervise AI documentation systems without additional compensation or training, managing workload that’s changed without their pay or job classifications changing, have a clear and articulable complaint. The legal support workers watching their departments shrink through attrition while their remaining workload increases have a clear and articulable complaint. The factory workers who are being monitored by AI performance management systems measuring things they weren’t told they’d be measured on have a clear and articulable complaint.
NLRB filings related to AI in the workplace increased 340% between 2022 and 2026. Most of these filings address AI-related grievances — unilateral deployment of AI monitoring without bargaining, use of AI to conduct work without notifying the union, AI-assisted performance management that workers argue constitutes surveillance beyond existing contract language. The frequency of these filings reflects genuine workplace anger, not just legal strategy.
The organizing wins in 2025-2026 with the clearest AI component are instructive. A successful union drive at a large insurance company’s claims processing center in 2025 was driven substantially by anger over the company’s rollout of AI claims automation: workers were told the AI was “helping” them, found that it was monitoring their productivity continuously and reporting to managers, and discovered that some of their colleagues were being denied employment offers (not laid off, but not replaced when they left) based on AI-generated performance scores that workers had no transparency into. The organizing drive message was explicit: “We don’t know what the machine says about us, and the company does.” That resonated.
Amazon warehouse workers’ union drives have had an AI dimension since at least 2021, when the monitoring and rate-setting algorithms became well-documented enough to serve as organizing rallying points. The pattern has continued and expanded. The algorithmic management critique — that workers are being directed, monitored, and disciplined by systems they can’t examine, appeal, or bargain with — is one of the cleaner labor rights arguments of the AI era, and it doesn’t require any particular political ideology to recognize as a legitimate workplace concern.
What’s not happening, or not happening fast enough, is organizing in the sectors most exposed to AI displacement: professional services, knowledge work, financial services. Union density in these sectors was already at rock bottom — 2-4% in many categories — and the workers experiencing the steepest displacement (mid-career knowledge workers in mid-tier cities) are not, historically or currently, organized. The cultural and institutional barriers to white-collar organizing in the US are substantial: professional identity often runs counter to collective identity, the workers most affected are geographically dispersed rather than concentrated in organizable workplaces, and the specific AI-driven employment reductions are happening through attrition rather than layoffs (which generates less visible anger and fewer organizing triggers).
The question of what labor power in an AI economy looks like — whether it continues to organize around traditional collective bargaining, or whether it develops new forms more suited to gig workers, displaced professionals, and the algorithmically managed — is genuinely open. The 2023-2026 period has produced some experiments. The Freelancers Union’s AI equity framework, which advocates for AI usage transparency and anti-substitution provisions in freelance contracts, is an interesting attempt to bring collective advocacy to a non-employment context. AI ethics boards with worker representation (mandated by California legislation in 2025 for firms above certain employment thresholds) are an institutional experiment in giving workers some visibility into how AI deployment decisions are made.
Neither of these approaches has the economic power of traditional collective bargaining. Both address something real. The historical parallel — again, the industrial revolution — is that labor power in new technological eras takes new forms that aren’t recognizable as the old forms until they’ve been established for long enough to seem obvious. The craft guilds of the pre-industrial era couldn’t be the primary vehicle for labor power in the industrial era. The industrial unions that replaced them in importance couldn’t have been imagined by a guild member in 1780.
What the 2026 labor relations data doesn’t tell us is what form of collective action will be adequate to the AI transition. What it does tell us is that the transition is producing the grievances and the shared interests that typically precede new forms of organizing. Whether those preconditions produce effective new institutions, or disperse into individual strategies of adaptation and exit, depends on leadership, political conditions, and contingencies that are not predictable from the current data.
The unions having the most impact in 2026 are the ones that understood early that AI wasn’t just a productivity tool but a governance question — a question about who gets to decide how work is done, who gets monitored and how, and who bears the cost when the decisions are wrong. Framing AI deployment as a labor rights question rather than a technology adoption question is not a rhetorical trick. It’s an accurate description of what AI deployment in the workplace actually involves. The organizations that figured this out three years ago are more relevant today than the ones that treated it as a separate issue from the core work of labor relations.
One email a week with practical Mac workflows, useful comparisons and honest tech recommendations.