The New Labor Movement: How Worker Power Is Being Rebuilt in the AI Decade

Section 01After Forty Years of Decline, Worker Power Is Rebuilding

In 1983, the first year the U.S. Bureau of Labor Statistics published its current union-membership series, 20.1% of U.S. wage and salary workers were union members. In 2024, the figure was 9.9% — under half [1]. Private-sector density is closer to 6%. By the conventional measure, the U.S. labor movement has been in continuous decline for four decades, and the trend line, smoothed, is still pointing down.

And yet anyone who watched the last four years closely knows the conventional measure is missing the story. The 2022–2024 window produced the largest single-firm unionization waves in a generation (Starbucks, Amazon, REI, Trader Joe's), the most consequential autoworker strike since the 1970s (the UAW Stand-Up Strike), the entertainment-industry strikes that put generative-AI provisions into a national collective-bargaining agreement (SAG-AFTRA and the WGA), and the first national platform-work regulation in a major economy (the EU Platform Work Directive). The U.S. labor-policy debate has not caught up. This piece is about what was actually built — sectoral bargaining, worker cooperatives, algorithmic-management regulation, portable benefits — and what 2026 worker power looks like in practice.

Section 02What the Last Four Years Actually Built

The five organizing and bargaining events of 2021–2024 that actually changed the policy landscape:

Starbucks Workers United

Beginning with a single Buffalo store in December 2021, Starbucks Workers United (affiliated with Workers United/SEIU) organized roughly 500 Starbucks locations in three years — the largest service-sector organizing wave since the 1930s. The campaign survived an aggressive union-avoidance response (which the NLRB found, in dozens of decisions, included unlawful retaliation), and culminated in a 2024 framework agreement between Starbucks and the union for a contractual path forward. The strategic lesson: store-by-store organizing using social-media-native communications, combined with relentless unfair-labor-practice filings, can move at speeds the 1990s playbook could not.

The UAW Stand-Up Strike (2023)

Under Shawn Fain, the United Auto Workers ran the first simultaneous strike against all three Detroit automakers in the union's history — using a rolling, plant-by-plant "stand-up" tactic that kept the companies guessing about where the next stoppage would come from. The settlements produced 25%+ wage gains over the contract life, elimination of multiple wage tiers, conversion of temporary workers to permanent, restoration of cost-of-living adjustments, and explicit provisions for organizing battery-plant joint ventures (the segment most exposed to the EV transition). The UAW has since used the contracts as the template for organizing drives at Volkswagen Chattanooga (won, 2024) and continuing campaigns at Mercedes Alabama and Tesla.

Amazon Labor Union — Staten Island (2022)

Chris Smalls and the Amazon Labor Union (ALU) won the first Amazon warehouse election in U.S. history at JFK8 in Staten Island in April 2022. The win was contested for two years; Amazon has refused to bargain. The ALU has since affiliated with the Teamsters, who are pursuing a national framework for Amazon organizing. The strategic question — whether the JFK8 win was a one-off or a template — is still open in 2026. What it definitively established is that Amazon is not unorganizable, the conventional industry wisdom for the prior decade.

SAG-AFTRA and the WGA (2023)

The 2023 SAG-AFTRA and WGA strikes produced the first major U.S. collective-bargaining agreements to address generative AI head-on. The contracts require studio disclosure when AI-generated material is used in writing rooms; require consent and compensation for the use of an actor's likeness to train generative models; and protect human writing credits. These are the first national contracts in any U.S. industry to embed AI worker protections. They are now the template every other union — and every employer counter-proposing — is starting from.

The EU Platform Work Directive (2024)

Adopted in 2024, the EU Platform Work Directive establishes a presumption of employment for platform workers (Uber, Deliveroo, Glovo and analogous platforms) subject to rebuttal by the platform, and creates the first transnational rules on algorithmic management in the workplace — including the right to human review of significant algorithmic decisions and disclosure of monitoring systems [2]. The directive is the most consequential platform-labor regulation in any major economy. Its implementation will shape what 2027–2030 platform work looks like across 450 million EU workers, and it will inform U.S. state-level proposals.

Section 03Sectoral Bargaining: The Model U.S. Labor Law Doesn't Currently Permit

The single largest structural reason for the gap between U.S. and Western European union density is not culture, not employer hostility, not "right-to-work" laws — though all of those matter. It is the U.S. system's restriction of collective bargaining almost entirely to the firm level (or, at most, the firm-plus-subcontractor level). In most of Western Europe, Australia, and parts of Latin America, collective agreements are negotiated at the sectoral or industry level — a single contract covering all employers in retail, all in metalworking, all in hospitality.

The mechanics: an industry-wide union bargains with an industry-wide employer association; the resulting wage scales, working-time rules, and apprenticeship structures bind every firm in the sector. The implication: employers compete on product, service, and productivity — not on the wages and conditions of their workers. A new firm entering the sector inherits the agreement; a workers' wage doesn't depend on whether their specific employer has been organized. Coverage rates in sectoral-bargaining systems regularly exceed 80%, even where union membership rates are far lower.

Kate Andrias and Brishen Rogers have argued in The New Labor Law (Yale Law Journal, 2016) and Data & Democracy at Work (MIT Press, 2023) respectively that the structural reform with the largest expected impact on U.S. worker power is enabling sectoral bargaining — not the legislative repair of enterprise-level rules (though that matters too) [3]. David Madland's Re-Union (Cornell ILR Press, 2021) lays out the operational design for a U.S. version of sectoral bargaining and surveys the state-level prototypes [4].

What U.S. sectoral prototypes look like in 2026

Three are worth knowing:

• California Fast Food Council (AB 257/AB 1228, 2022–2023). Establishes a council with worker, employer, and government representatives to set wages and conditions for the fast-food sector statewide. The 2024 minimum of $20/hour for the sector is the first effective sectoral wage floor in modern U.S. law.
• New York Nail Salon Industry standards. A combination of state regulation and industry agreement that emerged from the 2015 investigative reporting on the industry. Less formalized than California's model; produces sector-wide wage and safety norms.
• Minnesota Nursing Home Workforce Standards Board (2023). A council that sets minimum standards for the state's nursing-home workforce. Operates as the first U.S. care-sector standards board.

None are full sectoral bargaining in the European sense. All are tractable proofs-of-concept that the U.S. legal framework can accommodate sector-level standards-setting, where political coalitions exist.

Section 04Worker-Owned Cooperatives: A Small but Growing Stack

The U.S. has approximately 1,000 worker-owned cooperatives employing roughly 10,000 worker-owners (US Federation of Worker Cooperatives, 2024). It is a small share of the workforce. It is also approximately 4× the count in 2010 — a real growth trend, anchored in three things: the Baby Boomer business-owner retirement wave (worker buyouts of retiring owners), the post-2008 recovery of cooperative legitimacy, and a small but durable expansion of CDFI and impact-investor financing for conversions.

Worker cooperatives are not a substitute for unions. They are a different worker-power model: instead of bargaining with an employer, the workers are the employer. The empirical literature on co-ops (Pérotin, Ellerman, Logue) shows three durable patterns: cooperative firms have meaningfully lower productivity variance, modestly higher survival rates than comparable conventional firms, and substantially lower wage dispersion between executive and frontline workers [5].

Where the U.S. cooperative stack is actually growing

• Home care. Cooperative Home Care Associates (Bronx) is the largest worker cooperative in the U.S. with roughly 2,000 worker-owners, all home health aides. Several similar conversions have followed.
• Cleaning and facilities services. Dozens of immigrant-worker-led cooperatives operate in the New York, San Francisco, and Boston metros.
• Baby Boomer business succession. Project Equity, Democracy at Work Institute, and ICA Group facilitate worker buyouts of retiring small-business owners. The pipeline is real; the constraint is acquisition financing, which a small set of CDFIs (Local Enterprise Assistance Fund, The Working World, Shared Capital Cooperative) supply.
• Platform cooperatives. Driver-owned platforms (The Drivers Cooperative in NYC; Up & Go in cleaning) are early but operational alternatives to investor-owned platform incumbents.

The honest framing: U.S. cooperatives are a viable model at the edges, not the center. They matter as a worker-power option in the niches where they fit (succession, care, immigrant-led service work), and as a policy signal that worker ownership is a legible form of enterprise.

Section 05Algorithmic Management Regulation: The New Frontier

The largest unregulated category of workplace power in 2026 is algorithmic management — the automated systems that schedule shifts, allocate tasks, monitor productivity, score performance, set pay, and trigger discipline. Veena Dubal has documented (UC Irvine Law, multiple 2023–2024 papers) that algorithmic wage discrimination is now a primary mechanism of wage-setting in platform work, with materially different rates offered to identical workers for identical tasks based on opaque platform-side variables [6].

The regulatory frontier opening in 2024–2026 covers four areas:

1. Transparency mandates

The EU Platform Work Directive, New York City Local Law 144 (automated employment-decision tools), Illinois's AI Video Interview Act, and a growing list of state proposals require employer disclosure of algorithmic systems used in hiring, monitoring, and termination. Disclosure is the floor, not the ceiling — but it is a precondition for everything else.

2. Human-in-the-loop requirements

The EU directive establishes a right to human review of significant algorithmic decisions affecting workers. California's AB 2930 (proposed) and similar state-level proposals create a U.S. analog. The principle: certain employment decisions (termination, promotion, denial of work) cannot rest on algorithmic output alone.

3. Anti-wage-discrimination rules

Algorithmic wage discrimination — variable pay rates for identical work based on opaque worker-segmentation models — is the cutting-edge regulatory question. Seattle's PayUp ordinance for app-based gig work establishes a minimum-pay floor that constrains the most aggressive platform pricing. California Proposition 22's narrow exemptions for app-based work remain the largest counter-example. The 2026 question is whether the floor model spreads to other states, or whether the Prop 22 exemption model does.

4. Productivity-monitoring rules

Warehouse productivity quotas (Amazon's "rate" system most prominently) and white-collar productivity monitoring (keystroke tracking, screen recording, AI "engagement scoring") are emerging regulatory targets. California's AB 701 (2021) requires warehouse employers to disclose quotas and prohibits quotas that prevent rest-break compliance. Several other states have considered analogs.

Section 06What Doesn't Survive Contact with Reality

The rebuilding story is real. It is also incomplete, and there are pieces of the conventional worker-power narrative that don't survive a close look.

"Right-to-work" laws are decisive

The empirical record is mixed. Right-to-work states have lower union density on average, but the gap is not as large as the headline framing suggests, and the directionality of causation is contested. Several non-right-to-work states have similarly low private-sector density. The bigger predictors of organizing success are sectoral concentration, employer business model, and the strength of state-level NLRB-equivalent enforcement.

"Anti-strike rhetoric" doomed labor's standing

Public approval of unions in 2023 hit 71% (Gallup), the highest since 1965. The cultural narrative that workers turned against unions in the 1970s–1990s was real for that period and is no longer descriptive. The 2026 cultural environment is the most pro-union in two generations. The constraint is structural, not cultural.

The NLRB is functional

The National Labor Relations Board has been funded at roughly half its real-dollar 1979 level for the last decade. Election processing delays are routine; remedial orders take years to enforce; unfair-labor-practice case backlogs are at multi-decade highs. The Board's authority to order remedies remains limited by the Supreme Court's 1969 Gissel framework, which makes injunctive relief the exception rather than the default. EPI's annual State of Working America documents the operational degradation in detail [7].

Salting and union-avoidance consulting are minor

The union-avoidance industry — the consultants employers hire to run anti-union campaigns — is a multi-billion-dollar industry that the U.S. labor framework largely permits to operate without disclosure. The Form LM-10/LM-20 disclosure requirements were partially repealed in 2018; the Biden-era restoration was partial. Estimates from labor scholars place captive-audience meetings in the great majority of organizing campaigns and characterize them as a meaningful constraint on organizing success.

"Workers don't want unions anymore"

The most-replicated finding in workplace surveys is that approximately half of non-union U.S. workers would vote yes to union representation if an election were held at their workplace tomorrow (multiple replications, Hertel-Fernandez et al., 2020 through 2024). The gap between worker preference and actual union membership is one of the largest representation gaps in any major economy. The constraint is on the supply side (organizing capacity, employer opposition, NLRB underfunding), not the demand side.

Section 07What Workers Are Actually Doing in 2026

The 2026 worker-organizing landscape is more varied than the 1990s union-or-nothing binary suggests. Five forms of worker action are visibly operational right now.

1. Traditional NLRB-elected unions

Still the foundation. The 2022–2024 organizing wave was concentrated here. The constraints are documented above; the upside is that a contract negotiated under the NLRA is the most durable form of U.S. worker power.

2. Worker centers and minority unions

Organizations that don't seek NLRB recognition but operate as worker-voice institutions in low-density sectors. The Restaurant Opportunities Center, Fight for $15, the National Domestic Workers Alliance, and the Coalition of Immokalee Workers are the most prominent. They produce policy outcomes (state-level minimum wages, paid sick leave, the Florida tomato Fair Food Program) that NLRB-traditional unions in their sectors haven't.

3. Sectoral standards boards (where they exist)

The California Fast Food Council and Minnesota Nursing Home board are the early prototypes. Where the political conditions support them, they produce wage gains for workers whose employers are too small or fragmented to be organized firm-by-firm.

4. Worker-owned cooperatives

A small but expanding option, primarily in succession-conversion and immigrant-led service-sector contexts.

5. Independent worker associations

Slack-coordinated employee groups, professional associations, and informal collective-action networks that operate outside the NLRA framework — sometimes producing meaningful changes to workplace policy via collective public action (the Google walkouts, the Activision QA campaign, the Apple retail organizing) without ever filing for a formal election. These do not produce contracts, but they do produce policy and culture change.

Section 08What Policy Should Do, Concretely

The policy reforms with the strongest evidence base, ranked roughly by expected impact:

1. Enable sectoral bargaining. The single largest expected lever. Federal enabling legislation (the Workplace Democracy Act and successors have offered drafts; the Madland Re-Union framework is a more operationally complete model) would allow tripartite sector standards-setting in industries where the parties opt in. State-level expansion of the California Fast Food Council model can demonstrate before federal action is politically feasible.
2. Reform NLRA remedies. The PRO Act's strongest provisions — particularly card-check recognition for clear majorities, monetary penalties for unfair labor practices, and ban of captive-audience meetings — would materially change the cost-benefit calculus for employers running aggressive anti-union campaigns. Passage is not currently politically feasible at the federal level; state-level analogs (Illinois, California, New York) are doing some of the work.
3. Algorithmic-management transparency and human-in-the-loop standards. The EU Platform Work Directive is the template; U.S. state-level versions (California, Washington, New York, Illinois) are the near-term realistic path.
4. Portable benefits architectures. For workers in genuinely independent contracting relationships, benefit portability (health coverage, retirement, paid leave) decoupled from employer attachment is a structural reform with meaningful worker-welfare upside. The state-level prototypes (Washington's portable-benefits pilot; New York's portable-benefits framework discussions) are early.
5. NLRB funding restoration. The cheapest, most-impactful reform. Restoring real-dollar 1979 NLRB funding would meaningfully reduce case-processing delays and improve enforcement on the books, without changing a word of substantive law.

The fundamental problem of American labor law is not its substantive standards, which on paper compare reasonably with peer economies. It is its enforcement, its remedies, and the gap between formal rights and the practical capacity to exercise them. Brishen Rogers, Data and Democracy at Work (MIT Press, 2023)

Public approval of unions is at the highest level since 1965. Roughly half of non-union workers would vote yes to representation. The labor movement isn't out of demand. It's out of the legal infrastructure it needs to convert demand into membership.

The 2026 worker-rights conversation should not be a referendum on whether the 1950s industrial-union model is coming back. It is not. The honest question is whether the U.S. policy framework can accommodate the worker-power forms that are actually being built — sectoral standards boards, algorithmic-management rules, cooperative succession, hybrid unions-and-worker-centers — before the next decade of technological change makes the question moot.