AI-set insurance prices are acceptable to just 34% of European insurance consumers, who say they are comfortable with AI deciding a policy’s price without any human involvement, according to a Guidewire-commissioned Censuswide survey. The finding lands weeks before the EU AI Act reaches full applicability and months after EIOPA formally classified life and health pricing algorithms as high-risk — turning a consumer-sentiment poll into a real-world check on whether Brussels’ AI governance rules match what buyers actually demand.
The survey, fielded by Censuswide among 4,004 consumers across the UK, Spain, France and Germany between 13 and 22 January 2026, is part of an annual tracking exercise Guidewire has run since 2020, polling over 4,000 customers each year. Its 2026 wave arrives at an unusually pointed regulatory moment: insurers now have a hard compliance deadline to reconcile pricing-AI practice with supervisory expectations that were still theoretical when the survey series began.
A Third Say Yes, a Quarter Say Never
The topline split is stark. While a third of respondents accept fully automated pricing, 26% say nothing would make them confident in the insurance industry’s use of AI — a hard-skeptic bloc nearly as large as the acceptance camp. That leaves insurers with a majority of customers who are neither firmly opposed nor unconditionally comfortable, but conditional: their trust depends on specific design choices insurers make when deploying pricing algorithms.
Those conditions are not vague. Asked what would build confidence in an AI-driven decision they disagreed with, 39% of respondents ranked the ability to refer the decision to a human as their top confidence builder, well ahead of any other option. Explainability and external oversight tied for second place: 25% cited an explanation of why the AI reached its decision, and 25% cited independent regulation as the safeguards that would reassure them. Guidewire’s data, in other words, describes almost exactly the compliance architecture EIOPA has already mandated.
EIOPA’s High-Risk Label Already Anticipated the Trust Gap
Regulators got here first. In its Opinion on AI governance and risk management, EIOPA stated that AI systems used for risk assessment and pricing in life and health insurance are deemed high-risk under the AI Act. Notably, those specific systems are excluded from the Opinion’s own scope, since high-risk or prohibited AI already falls under the AI Act’s separate governance obligations — meaning life and health pricing algorithms face the strictest tier of EU rules, not a lighter insurance-sector carve-out.
The Opinion itself, addressed to national supervisors to clarify insurance-sector legal principles for the use and supervision of AI systems, sets out supervisory expectations that read like a direct answer to the survey’s confidence-builder rankings. EIOPA’s expectations span data governance, record-keeping, fairness, cyber security, explainability and human oversight — the explainability and human-oversight pillars mapping almost one-to-one onto the 39% “refer to a human” and 25% “explain the decision” responses. This is not coincidence so much as convergence: regulators and consumers have independently arrived at the same two non-negotiables for algorithmic pricing.
EIOPA has framed its approach as deliberately calibrated rather than blanket-restrictive. The authority describes its supervisory expectations as following a risk-based and proportionate approach aimed at promoting greater supervisory convergence among national competent authorities, rather than a uniform EU-wide rulebook applied identically regardless of a firm’s size or risk profile. That proportionality principle is precisely what insurers will need to invoke as the August 2026 deadline forces harder compliance choices, a point covered in InsuraBeat’s earlier analysis of EIOPA’s AI governance mandate and its penalty structure.
The August 2026 Clock Is Now Audible
Timing sharpens the stakes. Under the European Commission’s regulatory framework, the AI Act entered into force on 1 August 2024 and becomes fully applicable two years later, on 2 August 2026, with some exceptions, a deadline the Guidewire survey results now precede by roughly seven months. The Commission’s own regulatory framework for AI defines high-risk use cases as those that can pose serious risks to health, safety or fundamental rights — language that squarely captures pricing models determining what someone pays for life or health cover.
EIOPA has also signalled that fair treatment of consumers will stay under active supervisory watch beyond the AI-specific Opinion. Its 2026 strategic priorities list fair treatment of consumers in claims management, including through digitalisation, as a standing conduct focus area — a reminder that AI oversight will not be siloed from broader fair-value supervision, but treated as one strand of it.
The FCA’s GAP Insurance Files Show What Conduct Risk Looks Like
If EIOPA’s framework describes the rules, the UK’s Financial Conduct Authority has already shown what enforcement looks like when pricing outcomes fail a fairness test — with or without AI involved. In a fair value data release on guaranteed asset protection (GAP) insurance, the regulator found that for some products only 6% of the amount customers paid in premiums was returned in claims payouts, while some firms paid up to 70% of premium value out in commission to distributors such as motor dealerships. FCA Director of Insurance Matt Brewis said at the time: “Customers should be reassured that we’re in their corner and taking action where poor value is provided.”
That GAP episode functions as a template for how AI-set pricing could be judged going forward: not on whether an algorithm was involved, but on whether the resulting value distribution between customer and distributor holds up to regulatory scrutiny. An opaque AI pricing engine that produces a 6%-payout, 70%-commission split would fail a UK fair-value review regardless of how the price was calculated — which is exactly why explainability, not automation itself, is the axis both regulators and the 25% of skeptical consumers are focused on. Insurers already piloting consumer-facing AI, such as the AI-driven policy tools already live in France, will need to show that automation improves rather than obscures that value chain.
Mini-FAQ
What share of European consumers accept AI-set insurance prices?
Which AI insurance systems does EIOPA classify as high-risk?
When does the EU AI Act become fully applicable?
Sources
- EIOPA — Opinion on AI governance and risk management
- EIOPA — Union-wide strategic supervisory priorities and focus areas 2026
- Guidewire — European Insurance Consumer Survey 2026 (Censuswide)
- Financial Conduct Authority — Fair value data on GAP insurance
- European Commission — Regulatory framework for AI
