German Workers' Councils Win AI Audit and Shutdown Powers as Shadow-AI Risks Surge
13.06.2026 - 00:32:29 | boerse-global.de
Only 24 percent of organizations say their internal AI guidelines are consistently followed, according to a recent study from IT firm Ivanti. That governance gap—highlighted by the finding that 56 percent already deploy AI broadly in IT workflows—comes as Germany locks in a national legal framework for artificial intelligence and workers' representatives push for concrete rights to inspect, audit, and even turn off AI systems.
The German Bundestag passed the national implementation act for the EU AI Regulation on Friday, designating the Federal Network Agency (BNetzA) as the central market-surveillance authority. The agency will also serve as a public complaints office, advise companies, and run a regulatory sandbox to foster innovation. The coalition of CDU/CSU and SPD voted in favor; opposition lawmakers criticized what they called excessive bureaucracy and potential conflicts of interest.
One-time implementation costs are estimated at roughly four million euros, with recurring annual expenditures of 15.9 million euros for the federal government and 33.1 million euros for the Länder. The law provides for a two-stage evaluation procedure and fines of up to 50,000 euros.
Alongside the legislation, the Salzburg Chamber of Labour (Arbeiterkammer Salzburg) has released a practical guide for employee representatives. Author Wolfgang Goricnik supplies model wording for works agreements covering AI rollouts. The central demands include a works council’s right to shut down systems that violate rules, technical access to logfiles and training data, a ban on AI-driven performance and behavior monitoring, mandatory bias audits, and a prohibition of emotion recognition in the workplace. Early involvement of the works council is essential to avoid deadlocks and ensure transparency, experts stress.
A parallel threat is “shadow AI.” The Lower Saxony data protection authority (LfD Niedersachsen) warned Friday about employees using private AI accounts for professional tasks without official clearance. When staff fire up tools such as ChatGPT or Gemini without authorization, it creates significant data-protection and security risks. The LfD reminded employers that the EU AI Regulation obliges them to provide AI-skills training.
Beyond compliance, the technology is already reshaping staffing. The Ifo Institute published fresh figures Friday: nearly 20 percent of companies that use AI say it may become possible to replace skilled workers with less-qualified but AI-assisted personnel. The potential is seen as especially strong in retail and services.
A concrete example comes from the Upper Austrian Chamber of Commerce (WKOÖ). As part of a reform package, it aims to save around ten million euros by 2029, cutting 40 to 50 positions and offsetting the lost capacity with AI.
Microsoft Austria CEO Hermann Erlach described the development as an economic watershed. AI is no longer a pure IT matter, he said; it must be framed as a leadership and productivity issue. Infrastructure providers like CANCOM and HPE add that standardized data concepts are the technical foundation for successful deployment.
