Is ChatGPT an Insurance Distributor in the EU?

EIOPA Q&A 3407 asks the European Commission whether AI chatbots fall under the Insurance Distribution Directive (IDD). The answer could reshape how we think about AI in financial services.

Question 3407, submitted to EIOPA in August 2025 and now forwarded to the European Commission, asks whether publicly accessible AI chatbots fall within the scope of insurance distribution under the IDD.

I have been writing about this dynamic recently in light of the Nippon Life lawsuit against OpenAI. That was the US. But the question is now being raised formally in Europe as well.

Q&A 3407 builds on EIOPA's earlier Q&A 3168 on comparison websites. There, EIOPA concluded that when a comparison website requires a pre-purchasing questionnaire, asks for personal information like age, previous claims or medical condition, and then suggests specific products, that qualifies as insurance distribution. The new question asks the logical next step: does the same reasoning apply to AI chatbots?

The person who submitted it tested this in practice. They describe interactions where AI chatbots provided detailed offers for specific insurance products from specific companies, and in some cases recommended an appropriate level of coverage based on personal data and financial constraints. That is not a hypothetical. That is someone getting what looks like personalised insurance advice from ChatGPT. It is also worth noting how much technological capabilities have advanced since September.

What makes chatbots different from comparison websites is that they are not constrained by predefined forms or limited question sets. A consumer can share extensive personal information across multiple conversations. If prompted, chatbots can recommend specific insurance products from particular providers, including coverage details and estimated price ranges. As the Q&A puts it, "this interaction could resemble personalized advice and may influence the consumer's decision to purchase a specific insurance product, potentially qualifying as insurance distribution."

The question to the Commission is straightforward: could such chatbot interactions be considered as taking "additional steps to assist" in the conclusion of an insurance contract under Article 2(1)(1) of the IDD? And if so, what regulatory obligations arise for entities deploying such technologies?

The IDD was built for brokers, agents and bancassurance. It was written before the digitalisation era, and in my view it has navigated that transition quite well, largely because the definitions were drafted in a technology-neutral way. But it was not built for general-purpose AI tools used by hundreds of millions of people that can, in certain conversations, do what a regulated intermediary does, at least partly.

That raises some genuinely interesting questions.

I will be watching the Commission's response very closely and keep you updated here.