BaFin AI monitoring to cover banks and insurers under EU rules

BaFin AI monitoring will extend to banks, insurers and other financial firms that use artificial intelligence in regulated activities, Germany’s Federal Financial Supervisory Authority said Wednesday. The move follows new responsibilities for the watchdog under the European Union’s AI Act, adding another compliance track for financial companies adopting automated decision tools.

Jens Obermöller, BaFin’s director-general for cyber risks and technology in the financial sector, said in an interview published on the regulator’s website that BaFin will focus on AI used in connection with activities that already require authorisation, including banking and insurance transactions.

The regulator does not plan to review every AI tool at every institution. Obermöller said BaFin will examine samples of AI applications used across many financial firms in relevant areas, because the AI Act gives the authority a monitoring role rather than a broader supervisory mandate over all systems.

What will BaFin monitor in banks’ AI use?

BaFin will check whether firms comply with transparency requirements and rules on prohibited AI practices, according to Obermöller. The regulator will also look at whether companies are taking steps to improve employees’ AI literacy, a requirement aimed at ensuring staff understand the systems they deploy or oversee.

The AI Act is the EU’s framework for artificial intelligence, and BaFin’s role under it applies where AI use intersects with regulated financial services. In practice, that means the German watchdog will examine how financial firms use AI in authorised activities, rather than treating every internal experiment or general-purpose technology tool as part of the same review.

BaFin said its work will begin immediately, with a later focus on systems classed as “high-risk AI” from December 2027. Obermöller identified insurance systems used for risk assessment and pricing life and health insurance policies as examples of high-risk use cases.

In insurance, Obermöller said AI can be used to calculate individual surcharges payable by customers. For banks and other financial institutions, he said high-risk AI will generally include systems used to assess the creditworthiness and credit scores of natural persons.

The regulatory timetable comes as financial firms continue to test AI in lending, insurance and operations. In a separate PYMNTS interview earlier this month, Maik Taro Wehmeyer, co-founder and chief executive of Taktile, said he expected 2026 to be the year AI reaches financial services more broadly.

PYMNTS reported that AI could shorten decision times in areas such as commercial lending and insurance claims. The report cited small-business loans that previously required weeks of manual underwriting and could potentially be approved in minutes, and insurance claims that could be processed in hours with drone imagery and AI-supported damage assessments.

Wehmeyer said many people confuse AI transformation with cost savings, while arguing that faster decision-making may provide a larger competitive advantage. He gave the example of a small-business borrower receiving a loan decision in minutes rather than after a two-week wait.

This story draws on original reporting from PYMNTS.