AI Act and Insurance: What German Insurers Need to Know
Insurance is built on risk assessment—exactly what the AI Act regulates most heavily. When AI determines premiums, evaluates claims, or decides coverage, it’s making consequential decisions about people’s financial security. The EU classified this as high-risk.
For German insurers, AI Act compliance is not optional. It’s a core regulatory requirement.
Underwriting and Pricing Is High-Risk
AI that assesses risk and sets premiums for natural persons is explicitly high-risk under the AI Act. This includes automated underwriting, dynamic pricing models, risk scoring, and eligibility determination.
The requirements are comprehensive: risk management systems, data quality controls, bias testing, transparency to customers, human oversight, and ongoing monitoring. BaFin already oversees insurance, and they’ll coordinate AI Act enforcement.
Claims Processing Needs Attention
Claims AI lives in a nuanced space. Automated fraud detection that flags claims for human review is lower risk. But AI that denies claims or determines payout amounts makes consequential decisions—higher obligations apply.
The distinction matters: AI that assists humans vs. AI that decides. Structure your claims process accordingly.
Customer Communication
Chatbots and virtual assistants for customer service need transparency—customers must know they’re talking to AI. If these systems provide advice or make representations about coverage, accuracy and documentation become important.
What This Means Practically
German insurers need to map all AI touching customer decisions. Underwriting and pricing need full high-risk compliance. Claims AI needs careful classification based on decision authority. Customer-facing systems need transparency. Integration with existing BaFin compliance is essential.
How Compound Law Helps
- AI system inventory and classification
- Underwriting compliance frameworks
- Claims AI policy review
- BaFin integration strategy
- Bias testing and documentation
Frequently Asked Questions
Is all pricing AI high-risk? For natural persons, yes. AI determining premiums or eligibility is explicitly listed as high-risk.
What about commercial insurance? B2B insurance has lower obligations. High-risk classification focuses on natural persons.
Can we still use automated underwriting? Yes, but with full compliance: risk management, bias testing, transparency, human oversight, and documentation.
