German Public Sector Braces for EU AI Fines as Unions Demand a Seat at the Table
25.06.2026 - 03:13:50 | boerse-global.de
A new report from UNI Europa, released on 23 June 2026, warns that the gains from generative AI must be distributed fairly through social dialogue and collective bargaining. The same day, the ver.di union demanded a seat on Rhineland-Palatinate’s Transformation Council and called for the inclusion of the public service in the state’s official transformation agency. The union is pushing back against the use of AI as a mere cost-cutting tool. A follow-up event on AI in local government is scheduled for 14 September 2026.
The shift toward regulated AI adoption in Germany’s public sector is already underway. In February 2026, North Rhine-Westphalia issued a twelve-page code of conduct for the use of artificial intelligence in its ministries. The guidelines draw a clear red line: no personal or confidential data may be fed into AI systems. Permitted uses are largely confined to text-based tasks — drafting speeches, preparing presentations, composing emails, and polishing language. Explicitly banned are applications for social scoring, behavioural manipulation, and the monitoring of employee performance or conduct. The document stresses that final decisions must always remain under human control.
From 2 August 2026, those rules will be reinforced by the EU Artificial Intelligence Act. Transparency obligations kick in for deepfakes, and any company using AI in human resources — for screening job applications or monitoring staff, for example — must comply with new requirements. Violations can bring fines of up to €35 million. For systems that automatically assess learning outcomes, the EU AI Act classifies them as high-risk AI, with strict regulation set to take effect on 2 December 2027.
German works councils already have considerable leverage. Under Section 87, Paragraph 1, Number 6 of the Works Constitution Act, they enjoy co?determination rights over any surveillance equipment. The Berlin-Brandenburg State Labour Court (case reference 10 Sa 933/19) has previously underscored the importance of personality protection when digital tools are deployed.
Despite the regulatory hurdles, adoption is accelerating. A study by IDC commissioned by Dell Technologies found that 47.2 percent of German public authorities are already using generative AI, and a further 50 percent plan to do so. The biggest obstacles are data protection (cited by 66.7 percent), regulatory requirements (55.6 percent), and a shortage of skilled staff (52.8 percent). Meanwhile, the BMDS ministry has launched an “Agentic AI Hub” that is testing AI agents in 18 municipal projects to streamline administrative workflows — again with the caveat that full automation is off the table and human accountability must be maintained.
As worker representatives adopt digital stress matrices to assess the impact of new technologies, robust risk assessment becomes essential — both for psychosocial factors and physical workplace hazards. Having the right documentation in place helps you stay compliant and protect your team. A free toolkit with 41 ready-to-use templates, checklists and training materials makes it straightforward to get started. Download the free Risk Assessment Toolkit
Worker representatives are increasingly turning to digital stress matrices, based on a campaign by the European Agency for Safety and Health at Work (EU-OSHA), to assess risks. The urgency is clear: in 2024, mental disorders accounted for 16.7 percent of all days lost to incapacity for work. And roughly 40 percent of skilled employees say they are considering switching jobs because of concerns over AI.
