Artificial intelligence, better known as AI, has transformed the way we interact with data and technology, revolutionizing the way many industries use data to drive revenue and growth. The insurance industry, which relies heavily on data to market, underwrite and manage insurance products, is no exception to this technological movement. In particular, those involved in insurance marketing (often called “insurance agencies” or “insurance manufacturers”) see AI as a powerful tool for streamlining business operations and boosting sales productivity. However, the use of AI also poses a host of regulatory challenges which, in an industry that already has a well-established regulatory framework, could lead to significant additional regulatory burdens on insurance producers in the near future.
Use of AI in insurance sales
Using AI in insurance sales offers insurance manufacturers many benefits, similar to other sales-driven industries. For example, AI can automate mundane and repetitive tasks, such as insurance licensing and continuing education monitoring, data entry, and claims tracking, to increase productivity and reduce the time and costs associated with these functions. Additionally, chatbots can provide fast and accessible customer service 24/7, and AI-powered tools can help with risk management and fraud assessment. Generative AI can even help create marketing materials and blog posts. Overall, if used properly, AI can help insurance manufacturers operate more efficiently, reduce costs and avoid unnecessary human errors.
Pitfalls of AI
Despite its many benefits, regulators have indicated that the use of AI can lead to discrimination, lack of transparency and compromising data security, all of which are major issues in the insurance industry. While AI can help insurance companies process large amounts of data quickly, it can also lead to disparities in access to and pricing of insurance products. While AI can significantly reduce human error, AI itself is not without its flaws. For example, a Vectara study shows that chatbots ‘hallucinate’ false information between 3% and 27% of the time. See Simon Hughes, “Cut the Bull… Detecting Hallucinations in Large Language Models,” Vectara (November 6, 2023), https://vectara.com/blog/cut-the-bull-detecting-hallucinations-in-large- language models/. When processes are automated by AI, or decisions are made based on AI-driven interpretation of data, transparency in the data processing and decision-making process can be difficult to maintain, resulting in decisions and interpretations that are inaccurate or discriminatory. Additionally, the more data fed to a chatbot or generative AI system, the greater the risk of data leakage (i.e. accidental exposure of sensitive data or information).
Regulation of AI in the insurance sector
To date, regulation of AI in the insurance industry has focused primarily on insurance companies (i.e., the entities that assume the risk of the insurance product, also known as “insurers” or “underwriters”), and there is little regulation which effectively limits how insurance can operate. manufacturers may use AI in the marketing and sales of insurance. However, existing regulations for insurance companies around AI provide some guidance for insurance manufacturers considering implementing AI into their operations.
Model Bulletin of the National Association of Insurance Commissioners
In December 2023, the National Association of Insurance Commissioners (NAIC), a national organization of state insurance departments that works to standardize insurance regulations and is one of the industry’s leading authorities, adopted a model bulletin on the use of AI systems by insurers . NAIC Model Bulletin: Use of Artificial Intelligence Systems by Insurers, NAIC (December 2, 2023), http://content.naic.org/sites/default/files/inline-files/2023-12-4%20Model%20Bulletin_Adopted_0.pdf . Approximately eleven states have adopted a version of the Model Bulletin, and it is expected that more states will do so in the future. While the Model Bulletin and state-adopted orders do not carry the same weight as a statute or regulation, it can be largely helpful in understanding the way regulators think about AI and the direction the AI regulatory framework may take in the insurance industry .
Specifically, the Model Bulletin requires insurance companies to maintain a written program that ensures AI systems are used responsibly. Specifically, the program should focus on transparency, accountability and fairness, and hold insurance companies accountable for the use of AI systems developed by third parties by establishing document risk management and internal controls. Under the Model Bulletin, regulators are also given the authority to investigate the development and use of AI by insurance companies, including the creation and implementation of an insurance company’s written program for the use of AI.
Colorado law
Colorado is the first state to pass legislation regulating the use of AI in the insurance industry. Senate Bill 21-169 was signed into law on July 6, 2021, and prohibits insurance companies from using third-party consumer data and information sources, algorithms and predictive models unless the insurance company verifies or demonstrates that such use does not result in unfair practices. discrimination. In concrete terms, certain insurance companies are required to provide information on the use of external consumer data and information sources, and on the development and implementation of algorithms and predictive models, and to explain how the data, information, algorithms and models are used a establish and maintain a risk management framework, and provide an assessment of the results of the risk management framework for minimizing unfair discrimination.
However, before adopting such regulations, Colorado law requires the Colorado Insurance Commissioner to first hold meetings with stakeholders (e.g., insurance companies, insurance manufacturers, and consumer representatives) to adopt laws that appropriately include factors and processes relevant to certain types insurance. (such as life and private passenger cars). The Colorado Department of Regulatory Agencies, Division of Insurance (the Division) announced that it would first focus on the following areas: life insurance underwriting, private passenger auto insurance, and health insurance. See SB21-169—Protecting Consumers from Unfair Discrimination in Insurance Practices, Colorado Department of Regulatory Agencies, Division of Insurance, https://doi.colorado.gov/for-consumers/sb21-169-protecting-consumers-from-unfair- discrimination-in-insurance-practices (last accessed August 16, 2024). With respect to life insurance, the Division has issued regulations establishing governance and risk management requirements for life insurers using third-party consumer data and information sources, as well as algorithms and predictive models. See Color Code Reg. Section 702-10:10-1-1. The division is also currently holding meetings with stakeholders on a draft proposal for quantitative testing of algorithms and predictive models in life insurance. With regard to private car and health insurance, the Division recently held stakeholder meetings focusing on how the Life Insurance Regulation could be extended to these additional types of insurance.
Other states
While Colorado is the only state to date to adopt formal regulations specifically addressing the use of AI in insurance, other states, including California, New York and Pennsylvania, have recently considered bills that would also regulate the use of AI and insurance . Brian Joseph, “AI in Insurance: The Good, the Bad, and What Worries Regulators,” LexisNexis (December 11, 2023). As more states begin to adopt and implement the Model Bulletin, and Colorado continues to implement more comprehensive regulations, we expect more states to join the fray as regulators have the opportunity to assess the impact of the Model Bulletin and the Colorado – to investigate regulations on its use. of AI in the insurance sector.
What insurance companies should do
Even though the focus of current regulations is on insurance companies, insurance manufacturers are likely to feel the impact of the new AI compliance requirements imposed on their transportation partners. For example, if an insurer is required to establish and follow a written AI program aimed at mitigating negative consumer outcomes, insurance manufacturers working with that carrier will likely be required to comply with that same written AI program under the insurer’s supervision. Such programs will likely result in stricter risk management, data retention and data security policies for insurance manufacturers. Even though the Colorado regulations are aimed at insurers, they do limit how insurers can use third-party information and AI, with such restrictions likely trickling down to the insurer’s insurance producers.
Therefore, to stay abreast of the regulation of AI in the insurance industry, especially insurance sales, insurance manufacturers must ensure that any AI systems used are trained on unbiased data, and establish human checks and balances against the unavoidable AI. correct errors, maintain strong data security policies and systems, and maintain good record-keeping and transparency regarding the use of AI.
Jill Allison Opel, a partner of Foley & Lardner in the firm’s New York office, represents insurers and insurance-related entities across all industries, including casualty, life and health, property and casualty (including pet), surplus, travel and reinsurance. Margaret Brzakala is an associate in the company’s Milwaukee office. She helps insurers, manufacturers and other insurance-related entities achieve their business objectives while maintaining compliance with insurance laws. The authors would also like to acknowledge the assistance of summer associate Deajah Scott, JD Candidate University of Chicago Law School, Class of 2026.