A mid-June 2026 judgment from the European Court of Justice has thrown German human resources departments into uncertainty: unlawfully collected employee data does not automatically trigger a ban on using that evidence in court. National judges must now decide on a case-by-case basis whether the use of such data is proportionate.
The ruling arrives just weeks before the European Union’s AI Act imposes its first binding compliance obligations for high-risk systems on August 2, 2026. Companies that rely on artificial intelligence for hiring, performance reviews, or workforce analytics now face a dual squeeze — strict new rules on algorithms plus a shifting legal landscape on how employee data can be handled.
Penalties for violating the AI Act are steep: up to €35 million or 7% of worldwide annual turnover. Legal experts are urging immediate audits of every AI tool used in personnel management.
Shadow AI Adds Hidden Liability
One growing headache for compliance officers is so-called shadow AI — employees using free or public AI tools without official clearance. Without a formal risk assessment, such ad hoc usage can expose companies to both data-protection violations under the GDPR and violations of the upcoming AI Act.
The stakes are especially high in dismissals. Courts have accepted AI as a permissible aid in social selection procedures required under Germany’s Protection Against Unfair Dismissal Act, but only the employer can make the final decision. If the logic behind a scoring algorithm cannot be transparently reconstructed for employees or judges, any resulting termination becomes open to attack.
Works Councils Gain Leverage
Under Section 95(2a) of the Works Constitution Act, works councils have a mandatory co-determination right when companies introduce AI guidelines. Discrimination risks under the General Equal Treatment Act (AGG) and health-data protections under the GDPR must also be baked into the system from the start.
A concrete example of co-determination done right came on June 21, 2026, when the DGB Bayern awarded the MAN group works council for signing a company-wide agreement that enshrines the principle: AI should support workers, not replace them.
Preference for a Strategic Mix
The need for careful governance is underscored by a Bitkom study from May 2026, which found that 54.5% of German companies now use AI — a jump from 40.9% the previous year. Yet adoption varies sharply by sector: retail stands at 28.6%, while construction lags at 9.3%.
An Ifo study adds another twist: nearly one-fifth of AI-using companies believe the technology might allow lower-skilled workers to take over tasks previously reserved for university graduates. Market analysts note that strategies combining human oversight with AI tend to generate a clear net increase in headcount, while pure automation approaches yield smaller gains.
Workplace Surveillance in Crosshairs
Parallel to the AI Act, German labor courts are tightening limits on employer directives. The Düsseldorf Labor Court recently struck down a blanket four-day office mandate for IT staff. And Microsoft’s new “Workplace Check-in” feature, which can detect presence via Wi-Fi or IP addresses, now requires works council approval before deployment.
The message from recent rulings and the looming August deadline is consistent: technology is welcome as a tool, but employers cannot delegate judgment to a black box.
