€35 Million AI Fines Loom for German Employers as August Deadline and Union Push Collide
21.06.2026 - 01:31:54 | boerse-global.de
German companies that deploy artificial intelligence for hiring, performance monitoring, or workforce management face a legal reckoning from August 2, when the European Union’s AI Act imposes strict compliance rules for high-risk systems. Violators risk fines of up to €35 million or 7 percent of their global annual turnover—whichever is higher. The deadline is putting human-resources departments and works councils on a collision course with regulators, unions, and Germany’s already intricate labor laws.
Work councils have long played a gatekeeping role under the country’s Works Constitution Act. Section 87(1) No. 6 of the law mandates co-determination whenever an employer deploys technical tools capable of monitoring employee behavior or performance. The Federal Labor Court has clarified that even the potential to monitor—without proof of intent—triggers that right. Yet there are nuances. In early 2024, the Hamburg Labor Court ruled that simply using an open tool like ChatGPT without allowing the employer to view employee interactions does not require co-determination. Planning, however, is another matter: Section 90 of the same act gives the works council a consultative role during the design phase. When AI-based selection criteria are involved, Section 95(2a) demands explicit approval from the council. And the GDPR adds another layer: Article 22 bans fully automated decisions that produce legal effects.
Unions push for binding rules
On June 20, more than 140 worker representatives from IG Metall gathered in Duisburg to demand clear, enforceable standards for AI deployment. The message was that technology must be shaped as a tool serving employees, not the other way around.
A ver.di survey of over 8,000 workers in the insurance sector underscores why unions are turning up the heat. Eighty percent reported that AI has not reduced their workload. Nearly 90 percent saw no relief from job pressure. And more than 40 percent fear for their jobs—with the anxiety concentrated in lower income brackets. ver.di is currently negotiating a transformation-framework collective agreement with the employers’ association; the next round is scheduled for June 30 in Dortmund.
For HR departments, the August 2 deadline adds urgency. The EU AI Act categorizes personnel-management applications as high-risk, meaning they must meet strict transparency, accuracy, and human-oversight requirements. Experts advise businesses to conduct an immediate inventory of all AI tools in use and build a governance framework. That framework should include behavior-based access controls that dynamically evaluate activity and risk signals. In Germany, because such systems can monitor workers, they are also subject to works-council co-determination and typically require a data-protection impact assessment.
Collaboration outperforms automation
The Global AI Jobs Barometer 2026 from PwC offers a carrot alongside the stick: companies that pursue a combined strategy of human and AI collaboration saw workforce growth of 52 percent, while pure automation firms managed only 36 percent. Productivity among the human-plus-AI group reached 34 percent—more than double that of the automation-only group.
Still, the structural shift is painful. The Institute for Employment Research estimates that up to 1.6 million jobs could be affected by the AI transition, with roughly 800,000 disappearing and a similar number created. Some companies are already acting. Siemens is investing €500 million in AI-driven manufacturing in Erlangen. Ergo has announced plans to cut around 1,000 jobs, citing AI effects. Bosch intends to close its Waiblingen plant by 2028, affecting 560 employees.
Recent rulings by the Federal Labor Court have raised the bar for layoffs even further. Mistakes in mass-dismissal notices or filing a premature notification before completing consultations with the works council can render layoffs permanently invalid. For employers juggling AI implementation and restructuring, that means legal pitfalls demand as much attention as the technology itself.
