Michael von Gablenz: We’ve focused on three key areas:

• IP infringement mitigation: Our research shows that prompt-engineering techniques can reduce the similarity between generated images and training data, helping to lower infringement risks.
• Aggregated model risks: When multiple machine learning models interact, their errors can compound. We’ve quantified how correlations in algorithms or training data can amplify systemic failures.
• Risk assessment of machine learning algorithms: By applying conformal prediction, we can statistically bound the failure probability of predictive models, including multi-class classification algorithms, in high-stakes domains such as healthcare and engineering. This approach is model- and data-distribution agnostic, simple to implement, and provides reliable prediction intervals that help mitigate misclassification risk in practical use cases.

Each of these findings helps us design underwriting and risk quantification methods to appropriately cover new risks emerging from AI and GenAI.

Interestingly, while 74% of companies we surveyed are already using AI, about 60% of CEOs remain hesitant to invest further because of liability and ROI concerns. That hesitation is healthy - but it shouldn’t stop innovation. With the right safeguards and risk transfer solutions, companies can embrace AI with confidence. This is where our applied tools come in. At Munich Re Specialty, we’re actively putting research into practice by:

• Providing insight to help shape the evolving use of AI in areas such as law enforcement and public entity underwriting.
• Partnering with tech-savvy MGAs to unlock the full potential of AI in their operations.
• Collaborating with underwriters to refine policies and procedures so AI enhances — but doesn’t displace — human expertise.
• Developing pioneering solutions like aiSure™, which covers multiple models and loss scenarios created by AI errors, including lost revenue, business interruption, and legal damages.

These applications allow clients to innovate with AI while managing the risks in real time.

Michael von Gablenz: Don’t think of AI as just a technology challenge. Think of it as a risk management challenge. The organizations that will thrive are those that pair innovation with risk management - applying the same rigor to AI that they apply to financial controls or compliance. Do that, and AI becomes not just manageable, but transformative.

Munich Re family of companies have been at the forefront of insuring emerging technologies for decades. Our dedicated Insure AI team has been underwriting AI performance risks since 2018, drawing on deep expertise in statistics, mathematics, and advanced modeling. By combining rigorous research with practical industry experience, we help clients and partners worldwide navigate the uncertainties of AI adoption - designing innovative risk-transfer solutions that enable businesses to seize opportunity while managing exposure.