Firm counsels insurers on managing risk while using AI

Listen to this article

While AI technologies have revolutionized the insurance industry, it is not without regulatory, litigation, and reputational risk to those insurers. A quickly changing landscape on both the federal and state levels only makes these risks more complex.  

This month, the AI-X team at Faegre Drinker launched an algorithmic testing and AI governance and risk-management service for insurers. 

Insurers use AI models to help them make business decisions. These models can help them to better gauge risk and give customers tailor-made premium pricing. It can also help with them underwriting, claims decisions, fraud detection, utilization management, better marketing, and customer engagement. 

While AI has assisted insurers, as well as other businesses, in being more efficient and accurate, there are concerns about potential unfair discrimination that can arise from the use of AI models. These are concerns about accuracy, unintended discrimination against members of protected classes, and lack of transparency and explainability. This can open up insurers to significant risks. 

Law firms have jumped in to help companies mitigate those risks. One of those is Faegre Drinker. In April 2022, the firm announced an interdisciplinary artificial intelligence and algorithmic decision-making (AI-X) team. The team was created to advice clients in various industries on regulations and risk management stemming from AI technologies. 

It is composed of a deep bench of experienced attorneys who have advised clients on these issues for years as well as data scientists at Tritura. Tritura is a subsidiary of the firm, launched originally in 2014. QuarterJack, which is Tritura’s proprietary data analytics and AI platform, powers the testing.  

“Our clients are looking for help evaluating their processes against fast-paced regulatory changes in a way that smartly leverages technology,” Faegre Drinker partner and co-leader of the AI-X team Jay Brudz said in a news release. “We’re able to harness our QuarterJack platform to efficiently and effectively do just that.”  

Scott Kosnoff and Brudz, co-leaders of the AI-X team, help advise insurers on risks. In an email interview, they stated that, to mitigate the risks, “insurers should 1) stay on top of the evolving regulatory and litigation developments, 2) implement a risk-management framework that takes those developments into account and 3) consider testing for unfair discrimination, even if not legally required.” 

The AI-X team can help clients generate bespoke AI governance programs. “A risk-management framework should demonstrate that the organization appreciates the concerns associated with AI, takes those concerns seriously and has exercised reasonable care in identifying, managing and mitigating the risk of negative outcomes,” stated Kosnoff and Brudz. “Among other things, it’s important that risk management efforts be proportionate to the potential for harm and its likelihood of occurring.” 

Things are becoming complicated due to changing regulations and policies on both the state and federal levels. President Joe Biden recently issued a sweeping executive order regarding AI in October 2023.  

“We think the impacts on the insurance industry will be both direct and indirect,” stated Brudz and Kosnoff. “For example, insurers that do business with the federal government may be expected to adopt AI governance and risk-management practices that are consistent with those implemented by the federal agencies. In addition, technical guidance to be developed by the newly formed U.S. Artificial Intelligence Safety Institute (including on how to identify and mitigate algorithmic discrimination) could be embraced by one or more states in the insurance context.” 

States are also issuing governance regulation. In Colorado, for instance, a regulation was adopted in 2023 that requires life insurers who use external customer data and information sources (ECDIS) to utilize a governance and risk management framework that is designed to prevent unfair discrimination. Companies must comply with this by December 2024. Additionally, Colorado may adopt a regulation requiring life insurers who use ECDIS in the underwriting process to test to unfair discrimination. Life insurers may not be the only insurers affected; Colorado is looking at auto insurance and could consider other types of insurance. 

This month, the National Association of Insurance Commissioners (NAIC) unanimously adopted a model AI governance bulletin encouraging testing to identify potential “unfair discrimination in the decisions and outcomes resulting from the use of Predictive Models and AI Systems” and requires insurers to adopt AI governance and risk management framework. Many states may adopt this, or variations of it.  

“State-by-state differences are always a challenge,” said Kosnoff and Brudz in the email interview. “Fortunately, there’s a lot of overlap between the Colorado governance regulation and the just-approved NAIC AI model bulletin.”     

Insurers need to pay attention to these developments, advise Brudz and Kosnoff. “These state-based developments are a big deal and for many insurers will require a significant lift,” said Kosnoff in the news release. “Faegre Drinker is well-equipped to help our clients face these challenges and harness the power of AI in a safe, legal, and ethical manner.” 

“These regulatory developments are a big deal and for many insurers will require a significant lift,” stated Brudz and Kosnoff. “Putting together an effective risk management framework and testing for unfair discrimination aren’t the sort of things you can do over a long weekend.”