The role of artificial intelligence (AI) in the insurance industry has grown significantly, but its rapid adoption is accompanied by complex ethical, operational, and regulatory challenges.
Marc Voses (pictured), a partner at Clyde & Co, shared his insights into what insurance brokers in the United States should know about AI’s evolving regulatory landscape in 2025.
“The primary concerns regarding discrimination in AI underwriting – and the use of AI in claims handling or any operational capacity – are around bias in the algorithm and the data being used,” Voses said in conversation with Insurance Business. “AI systems can sometimes inadvertently perpetuate existing biases present in historical data.”
This bias, he says, can lead to unfair practices, including discriminatory pricing, coverage decisions, and claims handling.
“These concerns are pushing insurers to adopt robust fairness criteria, ensure transparency for consumers, and measure how AI systems are used to avoid disadvantaging any particular group,” he said. For insurance brokers, understanding these fairness criteria is critical to advising clients effectively.
AI promises to revolutionize underwriting by increasing efficiency and accuracy. However, Voses said that “the balancing act right now is between efficiency, accuracy, and the ethical challenges of using AI in the insurance process.”
Voses said that the industry needs to train AI on diverse datasets to minimize inherent biases and ensure continuous human oversight.
“AI isn’t going to create or manage itself,” he said. “You need human decision-making alongside AI to catch and correct potential biases.”
He also stressed the importance of transparency: “Regulators and consumers need to see how AI is developed and what data is being used.”
Continuous monitoring, he said, is equally vital: “Regular auditing and updating of AI models ensures they remain fair and accurate.” Brokers who are familiar with these practices will be better equipped to address client concerns about AI-driven underwriting.
Consumer skepticism remains a significant hurdle to AI adoption in insurance. “If consumers don’t believe a response is correct, insurers need to set up processes to challenge or appeal decisions,” Voses said. “If someone gets a response they don’t like, there must be an option to take that claim out of the AI system and have it handled by a human.”
He said that insurers must refine the customer experience by learning from past interactions. “Was it a communication issue? Did the consumer misunderstand the policy? Or did the insurer get it wrong? Learning from those experiences can improve the user experience and build trust, because this is absolutely the future,” he said.
For insurance brokers, this means reassuring clients that there are safeguards in place to support transparency and fairness in AI-driven decisions.
Voses predicted that AI’s efficiencies will lead to tangible benefits for consumers, such as reduced premiums and faster claims processing. “When consumers see that premiums are decreasing because processes are smoother, that’s a clear benefit,” he said.
“If someone gets insurance faster when they need it, that’s another benefit.” He also noted that regulation will also play a key role in building consumer confidence. “When regulators get involved, set processes, and reassure consumers that insurers are being held accountable, that builds confidence.”
“Consumers need to know regulators are safeguarding them and protecting the market,” he said. Brokers can play a pivotal role in educating their clients about these regulatory efforts.
Regulations are increasingly shaping the use of AI in insurance, particularly to address discrimination risks in underwriting. “At this point, I think it’s clear that regulation is essential for consumer acceptance and for guiding insurers to use AI effectively, efficiently, and ethically,” Voses said.
He believes regulators will prioritize bias audits, transparency requirements, and legal liability. “Regulators will want third parties to review insurers’ use of AI, including the software and datasets, to mitigate discriminatory outcomes,” he said. “They will require insurers to disclose AI usage and provide evidence that algorithms are free from bias.”
Legal liability frameworks, he noted, will ensure accountability: “A framework already exists to hold insurers accountable for any discriminatory outcomes caused by AI. This protects consumers and gives insurers clear guidelines.”
For insurance brokers, staying informed about these regulatory developments is essential to navigating the complex landscape of AI use.
In the United States, the National Association of Insurance Commissioners (NAIC) has already taken steps to regulate AI use. “The NAIC’s model bulletin ensures responsible AI use to mitigate adverse consumer outcomes and address governance, risk management, and internal audit functions,” Voses said.
He pointed out that over 20 states have adopted the framework, either as-is or with modifications.
“The framework focuses on transparency, accountability, consumer notice, and risk management. Insurers must assess risks, implement controls, and ensure privacy and data retention practices,” he said.
Additionally, he highlighted the importance of third-party vendor management: “External partners must also comply with the regulations.” For brokers, understanding how this framework affects their clients is key to providing relevant and timely advice.
AI has the potential to improve accessibility and fairness in insurance, provided it is implemented responsibly. “Ensure that AI systems improve coverage options and pricing while being fair and accessible to all consumers,” Voses said.
He underscored the need for responsible AI frameworks and enhanced consumer interaction. “Getting consumer buy-in is critical for AI adoption,” he said. By focusing on transparency, fairness, and trust, Voses believes the industry can navigate the challenges of AI adoption effectively.
What are your thoughts on this story? Please feel free to share your comments below.
