Navigating the AI Hype: Regulatory Challenges and Opportunities in Underwriting

The Gartner Hype Cycle and The Kennedys Global Report highlight the widespread discussions on AI in 2025, showcasing its risks and possibilities across industries. Despite the buzz, it remains uncertain, if not unlikely, that AI will become an effective tool in underwriting due to the stringent governance and controls required for effective risk selection.

A quick glimpse at regulatory approaches around the world will tell you that there is no common global approach to insurance regulation, nor is there an over-arching global AI philosophy, with different insurance jurisdictions have differing ideas about the risks and controls for AI. Couple that with global concern around wider ethical and moral issues, the potential for job displacement, misinformation and disinformation as well as the cultural adoption and education needed to truly embed AI into the workplace, it feels like it could be more effort than benefit. However it is happening, every firm is looking at how to leverage AI, so what are the regulatory considerations for Underwriting:

• Bias/discrimination: Addressing and mitigating biases in AI algorithms to prevent discriminatory practices
• Fairness in Underwriting: Ensuring AI-driven underwriting processes are fair and equitable for all applicants
• Transparency and Explainability: Making AI decisions understandable and transparent to stakeholders
• Privacy and Data Governance: Protecting customer data and ensuring compliance with privacy regulations

In the EU and UK, GDPR adds even tighter constraints under article 22, which sets out that the data subject shall have the right not to be subject to a decision based solely on automated processing, including profiling. The proposed Data Use and Access Bill may loosen the UK rules to allow more automated decision making in underwriting, however this is just one jurisdiction, with a significant legislative impact if not implemented correctly. Scaling globally, Data Protection is likely to create significant hurdles.

The reality is, AI in Underwriting is likely to be rules engine based, not true machine learning AI. Why? Firstly, all training data is likely to include bias, as reflected in the realities of the real world – this opens it up to likely risks of indirect discrimination both in training and in production variants, which (secondly) risks machine learning being inherently unfair to certain groups. Thirdly, where complex AI decision making has taken place, the AI may not be able to capture or explain its own decision logic fully at the time or post event.

Given the significant risks of AI and the major regulatory themes which are emerging, it leaves very little room for an effective way forward under a full AI and Machine Learning Framework. Further options including data augmentation under a human decision, or perhaps just a rules-based engine dressed up as AI also have potential regulatory implications.

Only 1 in 5 of the Gartner “big hype” events goes on to achieve mass market adoption, where AI falls is yet to be seen. What is important is any AI Underwriting proof of concept and/or implementation must involve regulatory and compliance considerations, not to stop you doing it, but to make sure you proceed with caution, actively considering the risks of AI throughout and making sure you get the right outcomes for your business and customers.

GreenKite’s experience with our clients suggest that as an industry, we are still in the early stages of exploring the possibilities of AI, with further work to explore possible use cases and to create an AI knowledge base within existing businesses. The risk and regulatory aspects of AI mean that as an industry we are continuing to be risk adverse, however AI is coming so it’s important the regulatory aspects and mitigation of risk should be at the core of all AI implementation discussions.

GreenKite support regulatory compliant business transformation strategies and solutions. We have in depth understanding of underwriting governance, assurance and automation through AI. If you want to be an early adopter and consider innovation through AI choose GreenKite as a partner with the regulatory and innovation expertise to set you apart from the competition.