The Chartered Insurance Institute has warned that a critical gap in artificial intelligence fluency across insurance and personal finance could undermine responsible adoption if firms deploy the technology faster than their people can understand, challenge and govern it.
The warning follows a roundtable the professional body convened in June, bringing together representatives from Chartered firms, technology providers, academia and other professional bodies to examine what good AI governance should look like across the sectors.
The resulting report, Responsible AI: from policy to practice, argued that boards, risk functions and customer-facing professionals need training that goes well beyond basic tool use, building critical thinking, professional scepticism and the confidence to question automated outputs.
The report's central argument is that this fluency gap will determine whether firms adopt AI with clear purpose and professional judgement, or end up driven by fear of falling behind competitors and a narrow focus on efficiency gains. It sets out specific behaviours the CII considers necessary for responsible adoption, including defining the problem AI is meant to solve before selecting a tool, and checking that any system is appropriate for the customer, business objective and level of risk involved.
Notably, the report pushes back on treating human oversight as an automatic safeguard. It warned against firms relying on human involvement alone, unless that person is active, informed and genuinely accountable for the outcome, a more demanding standard than simply having a person nominally review an AI output before it reaches a customer.
The CII's intervention comes against a backdrop of growing scrutiny of how UK regulators are handling AI risk. The House of Commons Treasury Committee concluded in January 2026 that the Financial Conduct Authority, Bank of England and HM Treasury were not doing enough to manage AI risks in financial services, criticising what it called a "wait-and-see" approach that could expose consumers and the wider system to serious harm.
The Committee found that more than 75% of UK financial services firms, particularly insurers and international banks, are already using AI, and recommended the FCA publish practical guidance on consumer protection and senior manager accountability for AI-related harm by the end of 2026.
That regulatory gap is precisely the space the CII's fluency argument is aimed at. If firms are moving ahead of clear regulatory guidance, the professional judgement and internal governance the CII is calling for becomes the main safeguard in the meantime, rather than a supplement to rules that don't yet exist in detailed form.
The fluency gap the CII described is not an isolated observation. A separate 2026 AI Impact Survey from Grant Thornton, covering 100 insurance executives, found that 44% said governance or compliance challenges have already contributed to AI project failure or underperformance, and only 24% said they were very confident their organisation could pass an independent AI governance review within 90 days.
The same survey found 39% of respondents identified frontline employees as needing the most support to adopt AI-enabled ways of working, and 29% cited talent or upskilling gaps as a top barrier to scaling AI. Coming from a source unconnected to the CII, that data points to the same underlying pattern: governance frameworks and AI deployment are outpacing the people expected to use them responsibly.
The concern is not hypothetical closer to home either. Insurance Business UK's own reporting has found 72% of surveyed Lloyd's market firms already have AI governance frameworks in place, with a further 21% developing them, yet deployment remains concentrated in operational efficiency rather than underwriting or claims decisions, suggesting governance structures are often outpacing genuine staff capability to use them well.
Separate reporting has highlighted that AI mandates without real capability-building tend to produce compliance rather than competence, with staff clicking through tools out of obligation rather than genuinely understanding them, precisely the dynamic the CII's fluency gap warning is describing.
Matthew Connell, director of policy and public affairs at the CII, said AI has the potential to improve how insurance and personal finance firms serve customers, but only if it is adopted with purpose, transparency and professional accountability at its core. He added that there is a responsibility to use these tools in ways that strengthen trust rather than dilute the human judgement and ethical standards that underpin the profession.
The CII has committed to supporting the development of practical AI policy frameworks, continuing professional development content, benchmarking tools, maturity models and live playbooks, while working with other professional and trade bodies on governance, procurement and standards.
The CII's argument is aimed broadly at firms and boards, but brokers have a specific, direct stake in whether it holds. When a carrier or MGA partner uses AI in underwriting, pricing or claims decisions, the practical question a broker should be asking is not simply whether a human reviews the output, but whether that person is genuinely equipped to challenge it: does the reviewer have visibility into how the model reached its conclusion, the authority to overrule it, and enough training to recognise when an automated decision is wrong for a specific client's circumstances. A carrier that can't answer that question with confidence is one where "human in the loop" may be closer to a formality than a genuine safeguard, and that's directly relevant to how confidently a broker can stand behind a placement or a claims outcome on a client's behalf.
What makes this warning notable is its timing relative to the regulatory picture. Firms cannot simply wait for the FCA's promised guidance before addressing internal capability, since the CII's argument is that fluency, not compliance paperwork, is what actually determines whether AI is used responsibly day to day.
A framework document or governance committee means little if the people using AI tools day to day cannot recognise when an output is wrong, incomplete or simply unsuitable for the customer in front of them.
That gap between having governance on paper and having genuinely capable staff is likely to become the more pressing test for firms over the next year, regardless of how quickly the regulatory guidance itself arrives.
