Sheldon Mills has a message for the British public. More than one in four UK consumers now place trust in tools like ChatGPT, Claude and Gemini when seeking financial direction. Yet few grasp a basic fact. Advice from these systems carries none of the safeguards that protect users of regulated advisers.
The Financial Conduct Authority’s executive director delivered that warning in comments that surfaced this week. And the implications stretch far beyond individual mistakes. Investing.com reported the FCA now plans a formal review within three to six months. The goal? Determine whether the regulatory perimeter must expand to cover general-purpose AI models that dispense investment tips, tax pointers or product recommendations.
This move arrives at a pivotal moment. The FCA published its long-awaited Mills Review on July 6, 2026. Led by Mills, the document maps how artificial intelligence could remake retail banking, insurance, investing and lending by 2030 and beyond. It builds on earlier work yet strikes a sharper tone on consumer risks.
Consumers already lean on these systems in surprising numbers. A Lloyds study found 28 million UK adults used AI for personal finance questions in 2025. Among those aged 18 to 34, 36 percent turned to the technology for investment ideas, according to Fidelity data cited by the Financial Times. That trust collides with performance that often falls short.
Which? tested leading chatbots last year. The results raised alarms. ChatGPT and Microsoft’s Copilot told users they could invest £25,000 in an Isa when the actual limit stood at £20,000. One prompt about tax triggered advice that would breach HMRC rules. ChatGPT insisted travel insurance was mandatory for most EU trips. Meta’s AI offered flawed guidance on claiming for delayed flights. Gemini suggested withholding payments from a builder, a step that could trigger a contract breach. The Guardian laid out the findings in detail.
Such errors matter. Unlike authorised advisers, these tools face no fiduciary duty. Consumers enjoy no access to the Financial Ombudsman Service or the Financial Services Compensation Scheme if things go wrong. The FCA has repeated that warning. “Unlike regulated advice provided by authorised firms, any advice provided by these general-purpose AI tools are not covered by the Financial Ombudsman Service and the Financial Services Compensation Scheme.”
Yet adoption keeps climbing. Some 81 percent of global financial firms now deploy AI in some form. Forty percent operate at advanced scale. UK institutions increasingly embed the technology in customer-facing roles. Chatbots handle complaints. They offer investment guidance. They explain product features. The pattern is clear.
FCA Chair Ashley Alder captured the regulator’s stance. “We need to keep pace with a rapidly changing environment and the principles-based, outcomes focussed approach we’ve taken on AI.” No rush to write pages of fresh rules. Instead, the emphasis falls on existing frameworks. The Consumer Duty. The Senior Managers and Certification Regime. Operational resilience standards.
But the Mills Review openly questions whether those tools suffice. It asks whether AI systems could deliver services “functionally equivalent to regulated activities while remaining outside the regulatory perimeter.” The paper floats scenarios where agentic AI — autonomous systems that act rather than merely advise — compare products, rebalance portfolios, even execute trades on a consumer’s behalf. Such capabilities could blur lines regulators once considered firm.
Risks extend past bad recommendations. Bias in training data can produce discriminatory outcomes. Opaque models make it hard to explain decisions. Concentration among a handful of technology providers creates new points of failure. One correlated breakdown across models could ripple through markets. Fraudsters already exploit deepfakes and synthetic identities at speeds that outpace traditional controls.
Opportunities exist too. Personalised guidance at low cost. Reduced consumer inertia through automated saving or switching. Greater participation in capital markets. The review seeks views on how the UK might harness these upsides while protecting those who lose out. Certain demographics may gain from tailored products. Others risk exclusion if algorithms deem them too risky or unprofitable.
Parliament has pressed for action. A Treasury Select Committee report urged the FCA to issue practical guidance by the end of 2026. Firms need clarity on how consumer protection rules apply to AI deployments. Senior managers require explicit expectations around accountability when harm traces back to an algorithm. The FCA’s perimeter report from March 2026 highlighted the same tension. Unregulated AI guidance sits at the edge of its remit. If consumer harm mounts, boundaries may need redrawing.
Recent coverage shows the debate accelerating. On the same day the Investing.com story appeared, City AM noted the FCA eyeing tougher rules as Britons embrace chatbots. Money Marketing reported the Mills Review predicts transformation in financial advice by 2030. Professional Adviser and FT Adviser have tracked calls for clearer guardrails throughout the year.
Industry experiments continue. The FCA runs AI Live Testing cohorts with banks and insurers. Barclays, Lloyds and others test models for credit scoring, anti-money laundering and customer support. Results will inform a good-and-poor-practice report later in 2026. The regulator’s AI Lab and Supercharged Sandbox, backed by NVIDIA, aim to accelerate safe innovation rather than stifle it.
Still, caution dominates consumer messaging. The Financial Times quoted experts who stress AI remains a tool, not an authority. One tax example illustrated the gap. A chatbot suggested a Monaco-friendly strategy useless for a Croydon employee because it ignored tapering allowances and prior contributions. “Big gap between ‘help me track spending’ and ‘tell me what stocks to buy,’” one adviser observed.
So what happens next? The FCA will digest feedback from the Mills Review, which closed in February but saw its full publication this month. Recommendations head to the board in coming weeks. A perimeter review could follow by year-end. Policymakers must weigh innovation against protection. Firms face pressure to document AI use under existing senior manager rules. Consumers, meanwhile, receive a consistent reminder. Check trusted sources. Do your own research. Treat chatbot output as a starting point, never the final word.
The tension feels familiar. Technology races ahead. Regulators follow with principles rather than prescriptions. This time the stakes involve not just individual losses but potential systemic vulnerabilities and eroded trust in financial markets. How the FCA threads that needle will shape Britain’s retail finance sector for the next decade. The review has begun. The hard choices lie ahead.
WebProNews is an iEntry Publication