Skip to content
You can now search across every topic, entity and event.What's new
AI: Jobs, Power & Money
17MAR

Refinery workers lose AI protection bid

3 min read
13:50UTC

Pittsburgh oil refinery workers with structural bargaining power demanded limits on AI surveillance and job guarantees. They received sub-inflation wages and no enforceable constraints.

← Back to Meta cuts 20% while Big Tech spends $650bnJump to analysis ↓
EconomicAssessed
Key takeaway

Labour's first AI monitoring test yielded no binding protections, confirming employers hold current bargaining leverage.

The United Steelworkers approved a bargaining programme in Pittsburgh covering more than 300 oil refinery workers, with demands including 25% wage increases and AI job protections — specifically, a block on management using AI to monitor workers' movements, assess productivity, and automate disciplinary decisions 1.

The outcome fell short on every front. Workers received sub-inflation wage increases and no binding guarantees against AI-driven job replacement. The monitoring restrictions — the most novel element of the demands — did not survive negotiations.

The result matters because these were not weak bargaining conditions. Oil refineries cannot be offshored. The workers hold specialised safety certifications that take years to obtain. The United Steelworkers is one of the largest and most experienced industrial unions in North America, with a decades-long record of extracting concessions from energy companies. If organised labour with structural leverage in a sector with high barriers to automation cannot secure meaningful AI protections, the prospects for less organised workforces are considerably worse.

Only roughly 10% of US workers belong to unions. The four strategies Labor Notes documented in March 2026 — monitoring restrictions, job guarantees, retraining provisions, and AI committee formation — are available only to that fraction. The New York Times tech workers' eight-day strike produced an AI impact committee but no binding job protections. Early AI-era bargaining is producing consultation mechanisms, not enforceable constraints. The gap between what unions are demanding and what they are winning is the gap the legislative proposals — the Warner-Hawley bill at federal level, California's SB 951 at state level — are meant to fill.

Deep Analysis

In plain English

The United Steelworkers union negotiated a contract for oil refinery workers in Pittsburgh. They wanted protection from AI — specifically, banning employers from using AI systems to track their movements, automatically evaluate their performance, or discipline and fire them via algorithm. They also wanted pay rises that kept pace with rising prices. They got neither. Wage increases fell below inflation, and there are no contractual limits on AI monitoring or automated job decisions. This is one of the first times a US union has explicitly tried to bargain over AI protections — and the failure sets a precedent that other unions will now face.

Deep Analysis
Synthesis

The USW outcome and the NYT NewsGuild result (Event 30) form a consistent pattern: employers concede procedural mechanisms — committees, oversight language — while protecting substantive economic interests — job guarantees, monitoring rights, licensing revenue. Procedural concessions are cheap and create the appearance of responsiveness. Economic concessions directly redistribute AI-generated value. Meaningful AI labour protections will require legislative intervention, not bargaining breakthroughs alone.

Root Causes

US labour law's 'effects bargaining' doctrine requires unions to negotiate the consequences of management technology decisions, not preconditions for them. Management retains unilateral authority to deploy new technology under most existing contracts; unions bargain only after deployment. This structural asymmetry — not negotiating strategy — explains the outcome and will recur systematically in subsequent AI bargaining cycles.

Escalation

The trajectory points toward more contested AI bargaining as monitoring tools proliferate across logistics, manufacturing, and services. Without a legislative floor equivalent to California's SB 951, outcomes will continue to reflect bargaining power rather than worker rights — and union density in oil refining has been declining for decades, weakening future leverage.

What could happen next?
  • Precedent

    The first explicit AI-protection bargaining outcome in US heavy industry sets a management-favourable template that subsequent union negotiations will face as a baseline expectation.

    Short term · Assessed
  • Risk

    Without binding job guarantees, refinery workers remain exposed to AI-based restructuring in the next contract cycle, with no contractual remedy yet established.

    Medium term · Assessed
  • Meaning

    The outcome confirms that bargaining leverage — not negotiating strategy — is the binding constraint on AI labour protections under current US law.

    Immediate · Assessed
  • Opportunity

    The clear failure of bargaining-only approaches may accelerate union support for legislative mandates such as SB 951 and S.3108, shifting the strategic focus from negotiations to lobbying.

    Short term · Suggested
Sources:Labor Notes
Mentions:Ukraine weapons export ban (2022) →Josh Hawley →Global South →United States →Iran-Azerbaijan Conflict (2026) →SB 951 →New York Times →Pittsburgh →Senator Mark Warner →Iran-Israel-US Conflict 2026 →
First Reported In

Update #1 · Meta cuts 20% while Big Tech spends $650bn

Labor Notes· 17 Mar 2026
Read original →
Different Perspectives
European Commission
European Commission
The European Commission's draft Annex III guidelines, closed for comment on 23 July, treat algorithmic scoring in recruitment, pay and termination as high-risk regardless of whether a human signs off, echoing Spain's Audiencia Nacional ruling 101/2026 on concealed scheduling algorithms. Brussels is shifting the fight from counting AI job losses to assigning legal liability for the tools themselves.
Office for National Statistics
Office for National Statistics
The Office for National Statistics recorded UK vacancies rising to 712,000 on 21 July, the first quarterly increase this beat has tracked, with payrolled employment down 85,000 on the year against May's 210,000 fall. The bulletin names no AI cause anywhere, and that is the point: nothing in the release confirms the displacement story it gets cited to support.
Christian Klein, SAP
Christian Klein, SAP
Christian Klein told investors on 23 July that SAP's research headcount will not grow for twelve months because AI agents and their token costs are absorbing the work, not because SAP is cutting jobs. He frames it as commercial arithmetic: the cost of AI-assisted coding tokens plus the salaries specialist AI hires command, not people being replaced by machines.
Betsey Stevenson, University of Michigan
Betsey Stevenson, University of Michigan
Betsey Stevenson argued that the 187,000 jobless-claims reading describes a market that hires little and fires little, not one AI is emptying. She said the real damage hides in eligibility rules and suppressed job postings, not in the headline layoff counts employers keep denying.
Comisiones Obreras, UGT and Concentrix's A Coruña works committee
Comisiones Obreras, UGT and Concentrix's A Coruña works committee
Comisiones Obreras, UGT and Concentrix's A Coruña works committee blamed Microsoft's push toward AI self-service for the 80 redundancies unions signed off on 22 July, not unavoidable business cause. A second Coruña procedure covering 80 more jobs runs to a 31 August deadline, and the unions want the state, not the employer, setting the pace of AI-driven cuts.
Stanford's 'We Must Act Now' signatories
Stanford's 'We Must Act Now' signatories
More than 200 academics, including 16 Nobel laureates, published a 13 July letter warning of AI-driven labour disruption, citing Daron Acemoglu's NBER estimate that AI's total factor productivity gain stays under 0.66% over ten years. The letter's own cited economics sit well below Goldman Sachs Research's 1.5-percentage-point estimate published the same week.