AI in banking in Germany

A customer chatbot is the AI use a bank's own customers notice first, and BaFin names it directly as an example of AI it supervises. Behind the scenes, transaction monitoring systems score payments for money laundering and fraud in real time, a task that section 25h(2) of the Kreditwesengesetz (KWG) already requires German banks to run, whether they build the system themselves or buy it as regtech. Credit scoring is the use with the most regulatory weight: a model that sets or affects the creditworthiness of a natural person counts as high-risk under the EU AI Act, while a model that only rates a company's credit does not fall under that category.

Large German banks such as Deutsche Bank, Commerzbank and DZ Bank, the central institution for Germany's cooperative banks, run their own data science and AI teams for these tasks, next to the model risk management groups that validate scoring models for capital purposes and now for AI compliance as well. Savings banks and cooperative banks, the largest part of the German retail market by branch count, mostly buy these systems from their central IT providers instead of building them. The monitoring duty under KWG section 25h stays with the bank either way.

BaFin supervises AI that banks use in direct connection with regulated financial activities under the German Act Implementing the European Artificial Intelligence Act, and names customer chatbots and creditworthiness assessments as examples in scope. Before that act, BaFin had already set out its own principles for the use of algorithms in decision-making processes, which ask every bank to name a clear owner for a model's development phase and its application phase. BaFin President Mark Branson summed up the aim this way: "People have to be able to trust that their fundamental rights will be protected when AI is used."

The Deutsche Bundesbank supports supervision of the largest banks together with the ECB and runs the German side of on-site checks. For a bank that scores consumer credit with AI, the deployer duties under the EU AI Act sit on top of the existing KWG monitoring duty, and a bought scoring system also becomes an ICT contract that belongs in the DORA register of information.

AI banking jobs in Germany sit inside the banks named above, at their IT subsidiaries, and at the consultancies and software vendors that build scoring, monitoring and chatbot systems for them. The first task most teams face is an inventory: listing every AI system in use and sorting it into prohibited, high-risk, transparency-only or minimal risk, since the line between a fraud model and a credit scoring model is not always obvious. The second is documentation: model risk management already validates scoring models for capital purposes, and it now adds the logging, human oversight and bias checks the EU AI Act asks for on top.

Generative AI has moved into the daily work of many banking teams faster than any earlier wave, mostly for drafting text, summarizing documents and searching internal knowledge, always with a person reviewing the output before it goes out.

Finance Loop is the meeting place for model validators, compliance officers and engineers who put AI to work inside German banks. It connects the finance, IT and AI communities in Frankfurt and holds events in Munich, Berlin and Hamburg too.

Finance Loop highlighted fAInance, a conference by Sopra Steria and Fraunhofer IAIS whose demo areas covered AI in software testing, governance, risk and compliance and AI against financial crime, aimed at staff of financial institutions. It announced KI Exchange 2026 in Hamburg, where AI governance and DORA compliance were on the program with speakers expected from BaFin, and it has a cooperation with Frankfurt Data Science on AI governance in financial institutions. Related pages: AI in finance in Germany, the EU AI Act in finance and regtech in Germany.

Customer chatbots, fraud and money laundering monitoring on payments, and credit scoring are the three uses BaFin names most often. Generative AI adds drafting, summarizing and internal search on top, with a person checking the result before it reaches a customer or a decision.

Yes, when it concerns a natural person. The EU AI Act lists AI that assesses the creditworthiness of natural persons or sets their credit score as high-risk, which means the bank must keep human oversight, monitor the system and assess its effect on fundamental rights before first use. Fraud detection and the scoring of companies are not covered by this category.

Yes. Under the German Act Implementing the EU AI Act, BaFin supervises AI that a bank uses in direct connection with regulated financial activities, and it names customer chatbots as one of the examples in scope.