ChatGPT at Work: What German Law Says About AI and Dismissal

German employers are issuing warnings and threatening dismissal over AI tool use. The law is still catching up, but some clear lines are emerging. What expats working in Germany need to know.

---

This is new territory, for everyone.

AI tools like ChatGPT, Claude, and Gemini are now part of how a lot of people work. German employers have started reacting: blanket bans, warnings, and in some cases dismissal threats. German courts are only beginning to work through what’s actually allowed.

What follows is the current state of play. It will keep developing.

What an employer can do

An employer can prohibit specific AI tools by workplace instruction (Weisung). That instruction has to be clearly communicated and documented. If you then use a banned tool, you risk a formal warning, and after that, dismissal.

What employers cannot do: have an AI system automatically decide to dismiss someone. Any termination decision has to have a human being behind it. Automated dismissals aren’t valid under German law.

What the real risk usually is

The most concrete legal exposure from AI tool use isn’t the tool itself, it’s data protection. If you enter customer data, internal business figures, personal data, or anything covered by professional confidentiality (especially relevant for lawyers, doctors, accountants) into a public AI system, that’s a potential breach regardless of whether there’s an official ban.

Worth noting: some employers have rolled out Microsoft 365 with Copilot while simultaneously threatening staff for using ChatGPT. Whether there’s a meaningful legal distinction between the two, in terms of where company data actually ends up, is a legitimate question that courts haven’t fully answered yet.

Where things are genuinely uncertain

If there’s no explicit ban and no confidential data involved, using AI as a work aid is hard to discipline. A Munich court (ArbG München) declined to uphold a summary dismissal against a journalist whose publication produced AI-generated content, partly on the basis that the boundaries of acceptable practice hadn’t been made clear in advance.

The principle that applies is the same one behind the Abmahnung rule: before you can be dismissed for conduct, the conduct has to have been clearly identified as prohibited. A vague culture of AI scepticism isn’t a valid workplace instruction.

What to do if you’ve received a warning for AI use

The first question is whether the prohibition was actually communicated to you, clearly, specifically, and before the alleged breach. A policy nobody announced, buried in an intranet nobody reads, or issued after the fact doesn’t meet that bar.

If you’ve received an Abmahnung over AI use, don’t simply accept it. A warning in your personnel file is the legal precondition for a dismissal. If it’s based on a shaky prohibition, it’s worth challenging.

FAQ

My employer just issued a company-wide AI ban. Does it apply immediately? Yes, from the point it was properly communicated to you. Violating it going forward puts you at risk of a formal warning.

Can my employer actually tell whether I used AI? Detection tools exist but are unreliable and carry little evidential weight in court. That said, if your employer knows, the trust is often gone regardless of what a court would say.

I used AI to help with a task, but the work was still mine. Is that a problem? Without a clear ban, probably not. Using AI as a tool rather than fully outsourcing the work is a relevant distinction, though courts are still working out exactly where the line sits.

I’ve received a warning. Is it worth challenging? Depends on whether the prohibition was clearly communicated beforehand. Get in touch with the details.

Contact: