🔇 The Silent Disarmament

How Automation Is Undermining the Right to Strike. What This Means for Employers, Unions, and Society

When the GDL paralyzed Deutsche Bahn with six strike waves in winter 2023/24, and when Hamburg port workers repeatedly shut down terminals in 2024, the public debate was predictable: right to strike, proportionality, societal damage. What barely came up in this debate was the actually relevant question. Not: Are they allowed to? But: What are they triggering?

DB's Head of HR Martin Seiler answered this question after the GDL settlement so clearly it was hard to miss: automation would now move more prominently to the foreground. Not as a threat. As a consequence. What Seiler described is not a German phenomenon and not a temporary reflex. It is a structural mechanism that is just getting underway. I call it the silent disarmament, because it doesn't announce itself in press releases but unfolds in investment decisions that are rarely publicly justified.

Thesis 1: The Employee Market Was a Historic Window. It Is Closing.

Between 2021 and 2022, the negotiating position of employees was stronger than it had been in decades. Job openings at record levels, skilled labor shortages in nearly every sector, companies courting employees. This was not the new normal. It was a demographic and economic window that opened quickly and narrowed just as fast.

Open positions have nearly halved since their peak. What follows is not a classic economic cycle. It is something more structural. Every position that becomes vacant today through retirement, resignation, or restructuring poses a decision question: Will it be refilled, or will the task be automated? This question was previously almost always answered in favor of humans, because automation was expensive, slow, and error-prone. That premise no longer holds.

Every vacant position is now an investment question. And the answer is increasingly different from before.

Thesis 2: The Logic Is Simple: What I Can Automate, I Don't Need to Staff. And What I Don't Staff Can't Strike.

I want to describe this calculation without sugarcoating, because it is real. An employer choosing between a position with labor costs, social contributions, vacation entitlements, sick days, possible collective bargaining and strike risks on one hand, and a system with investment costs that runs around the clock, makes no demands and knows no union on the other, will increasingly make this assessment pragmatically. The return on investment for automation has shortened dramatically in recent years. After the UAW strike in the US, it halved in the automotive industry to two years. In Germany, HHLA shows what this means concretely: fully automated terminals, remote-controlled cranes, driverless transport vehicles. No human who could strike.

There is also the standstill option. A company doesn't have to actively automate. It can simply wait. When a position becomes vacant and is not refilled, this doesn't trigger a business change as long as thresholds are not exceeded. Silent reduction through natural attrition is legally fully permissible and works council-free. It generates no headlines and no bargaining pressure. It is the silent version of the power shift.

Thesis 3: The Works Constitution Act Protects the Workplace, Not the Work. This Gap Has Explosive Power.

There is a legal asymmetry that barely appears in public debate. When a workplace is changed by AI, meaning parts of the activity are automated but the human remains, the works council has co-determination rights. The Works Constitution Act protects here. Since AI systems are almost always capable of performance and behavior monitoring, almost every AI introduction requires co-determination.

However, when a position is completely eliminated because an activity is fully automated, that is a fundamental business decision, and it is co-determination-free. The works council can negotiate the consequences. Not the decision itself. And when positions simply aren't refilled, the works council isn't even at the table, because no table is set. No one is dismissed. No social plan is due. The workforce shrinks silently.

I consider this the most important institutional gap in the entire automation debate. It wasn't created by legislative error but was always there. It gains a new quality through the speed and cost efficiency of today's AI systems: what used to take years and cost millions can now be implemented in weeks. The protective instruments were built for a different speed.

The works council can negotiate the consequences. Not the decision. This gap gains a new quality through the speed of automation.

Thesis 4: Partial Automation Creates Transparency That Some Didn't Want.

There is another dimension of partial automation that rightly concerns works councils. When AI systems take over parts of an activity, data points emerge as a side effect: How long does who take for which task? Where does performance deviate from the average? What serves productivity measurement simultaneously serves behavior monitoring. This is documented at Amazon warehouses, in ride services through app tracking, in office environments through Microsoft productivity tools. Courts draw red lines, but the absence of a standalone employee data protection law leaves considerable gray areas.

I consider this question one of the most underestimated in the entire AI debate. Not because surveillance is necessarily the purpose, but because the data emerges whether you want to use it or not. And because companies that don't use it may face competitive disadvantages against those that do. This is a structural temptation, not personal misconduct.

Thesis 5: The Public Sector Is Not a Safe Space Either. It Is the Next Chapter.

There is a widespread assumption that automation and labor conflicts are a private sector topic. The public sector, intuition suggests, is structurally protected by civil service law, collective agreements, and political consideration. This assumption does not withstand sober examination.

Hamburg's U5 will be Germany's first fully automatic subway line running without drivers. This is not the distant future but planned operations from 2029. Hochbahn is testing autonomous buses in regular city transport. McKinsey estimates that generative AI could close a third of the skilled labor gap in the public sector. Around 570,000 positions are currently unfilled there. The demographic pressure forcing the private sector to do more with fewer people hits public administrations and utilities with the same force, just with a delay. That delay is shrinking.

For unions like ver.di, this presents a strategic challenge. The strike works as leverage as long as vehicles need drivers and agencies need people. When neither is true anymore, the basis for negotiation changes. Not as intention, but as consequence.

Thesis 6: This Is Not Cause for Dystopia. But an Urgent Reason to Rethink Work.

I want to make a distinction that is almost never made in this debate. The power shift I describe is not an attack on employee rights. It is a structural consequence of the convergence of technological progress, demographic change, and rising labor costs. This changes the rules of the game without making the players evil.

The institution of the works council has not lost its historical legitimacy. It has gained a new task: not to block change, but to negotiate its conditions. The collective agreement that ver.di concluded with insurer Ergo in 2026 shows what this can look like: employment security until 2030, qualification entitlements, social transition agreement. This is shaping rather than managing.

What is missing is a societal debate about what work means beyond the employment relationship. Marie Jahoda empirically showed in the 1930s that employment fulfills functions that have nothing to do with salary: time structure, belonging, the feeling of being needed. Those who lose these functions through automation lose more than an income, even if the income is replaced. This question is not a footnote in social policy. It is the actual topic.

The institution of the works council has not lost its historical legitimacy. It has gained a new task: not to block change, but to negotiate its conditions.

The silent disarmament unfolds without announcement, in investment calculations and hiring decisions that rarely make headlines. Those who ignore it will be surprised by it. Those who understand it can help shape it, on both sides of the negotiation table.

Sources

1. IAB: Labor Demand Monitor 2/2025 and Job Survey 3/2025.
2. Federal Employment Agency: Year in Review 2025.
3. WSI / Hans Boeckler Foundation: Industrial Action Report 2024.
4. Acemoglu, D. / Restrepo, P.: Robots and Jobs. Journal of Political Economy, 2020.
5. NBER Working Paper: Minimum Wages and the Rise of the Robots. 2025.
6. Bird & Bird: First Ruling on Works Council Rights in AI Deployment. Hamburg Labor Court, Jan. 2024.
7. McKinsey: Generative AI and Talent in the Public Sector. July 2024.
8. ver.di / DGB: Ergo AI Transformation Collective Agreement. February 2026.
9. Hamburger Hochbahn: Project U-Bahn100 and ALIKE.
10. Jahoda, M.: Employment and Unemployment. Cambridge University Press, 1982.