Copy article

If you can’t explain it to your gran, you shouldn't be letting an AI do it with your client's money.

ended 21. January 2026

The Treasury Committee has hit the nail on the head: regulators are sleepwalking, and businesses are treating AI like magic pixie dust rather than a tool. The report shows 75% of firms are using AI, but nearly half admit they don't fully understand the tech they've plugged into their business.

In the advice world, we talk about "knowing your client." But if you’re using AI tools to sort data, spot fraud, or draft reports, you also need to "know your code." If you use a "black box" system that you can't explain, you aren't being efficient, you're gambling.

We want IFAs to use AI. It’s brilliant for handling the boring admin so you can spend more time face-to-face with clients. But there is a right way and a wrong way. The wrong way is trusting a vendor who says, "don't worry, the AI handles it." The right way is treating AI like a junior member of staff: you check their work, you understand how they got to their answer, and you never let them make final decisions on serious money matters without a human signing it off.

We'd like your views:

  • If an AI tool makes a mistake on a client file, are you ready to take the blame personally, or will you blame the software provider?
  • Is it fair to use "black box" tools for compliance if you can't explain to the FCA how the tool actually works?
  • How do we stop AI from becoming “automation theatre”, where it looks impressive but actually creates more mess for humans to fix?
  • Should every financial advice firm have a "Human in the Loop" rule, where no client communication goes out without a real person checking it?
  • Are we in danger of losing the "personal" in personal finance by automating too much of the relationship?

Treasury Report - AI in financial services

6 responses from the Newspage community

Copy all

Star Quote
Copy

Rohit Parmar-Mistry
Founder at Pattrn Data
The regulators are sleepwalking into a crisis, and the Treasury Committee has just sounded the alarm. It is staggering that while 75% of the sector uses AI, we still have no AI-specific stress testing. We test banks for market crashes, but we aren't testing them for when a critical algorithm starts herding the market off a cliff.

The report highlights a reliance on a tiny group of third-party providers, and the risk isn't theoretical, it's already here. The report points to the AWS outage in October 2025, but we also saw providers like Cloudflare go down multiple times in 2025. When these central nodes fail, they don't just take down a website; they freeze the financial system

FCA guidance by "end of 2026" is too slow. Technology moves in weeks, not years. By the time that guidance is published, the damage will already be done. We don't need more "wait-and-see"; we need practical, standards on transparency today. If you can't explain it, you can't sell it. Simple as that.
Copy

Editor
Tony Redondo
Founder at Cosmos Currency Exchange
Businesses must overcome "shiny new toy" syndrome and stop treating AI as magic pixie dust. Under the Senior Managers and Certification Regime (SM&CR), accountability cannot be outsourced; if the AI fails, you—not the vendor—are on the hook. Using "black box" tools is a breach of the Consumer Duty. If you cannot explain the logic behind a recommendation to the FCA, you have failed your professional obligation. To avoid "automation theatre," treat AI like a junior staffer: it drafts, but a human signs off. A "Human in the Loop" rule is essential to prevent systemic errors and preserve the empathy central to financial advice. As AI commoditizes data, the "personal" element becomes your only true premium. Automating the relationship risks automating your own relevance. In an era of algorithmic advice, you must know your client and know your code.
Copy

Editor
Rob Mansfield
Independent Financial Advisor at Rootes Wealth Management
With it being called artifical intelligence, it gives the impression of it being sentient, but it's just a fancy computer in reality. It's a tool, just the same as a calculator is and it needs to be checked. I dabble with AI without giving it client data and it's useful but I wont be risking my professional reputation on it at this stage. The pace of development is fascinating and so it's only likely to get better and then we'll see.
Copy

Kundan Bhaduri
Entrepreneur, Investor and Landlord at The Kushman Group
I have been fervently of the opinion that more than half of the IFAs in the market are not worth their cost. Many do not even understand what they are selling in the first place. What we are now witnessing is the industrialisation of laziness where small and mid sized financial firms are plugging black box models into their decision making processes without the faintest idea of how that engine works.

The recent statistic that nearly half of firms using AI cannot fully explain its outputs is not just a compliance failure, I would say it is a dereliction of their fiduciary duty.

If we lose the 'personal touch' in personal finance we are not innovating. We are instead simply commoditising the industry's obsolescence. The smartest firms in 2026 will be the ones that use AI to empower their well-trained advisors rather than replace them because eventually when the market crashes that LLM chatbot will not be there to hold your hand.
Copy

Patricia McGirr
Founder at Repossession Rescue Network
If an adviser cannot explain an AI decision, they cannot defend it. The Treasury Committee has exposed a quiet failure of governance. Firms are rolling out AI they do not properly understand and calling it efficiency. In financial advice, that is not progress, it is risk dressed up as innovation. When AI sorts data, flags risk or drafts client material, responsibility does not disappear. It hardens around the human who signed it off. AI is valuable when it removes admin and frees advisers to focus on clients. It becomes dangerous when it is treated as a black box authority. Any system influencing real money should be supervised like a junior employee, checked, questioned and never allowed to act alone. If a firm cannot clearly explain an AI outcome to the Financial Conduct Authority or a client, it should not be using it. Automation without accountability is not efficiency, it is abdication of responsibility.
Copy

Colette Mason
AI Ethics Consultant at Clever Clogs AI
AI in the advice world has become a double-edged sword. On one hand, the technology promises to automate administrative tasks, saving valuable time for IFAs to focus on client relationships. On the other hand, it is essential that AI’s role is not overestimated or blindly trusted.
The Treasury Committee is right to highlight the regulators' failure to understand the tech deeply and its widespread use without full comprehension of its limitations.

In the financial services sector, particularly in areas like suitability ranking, fraud detection, and recommendation report drafting, the emphasis on "knowing your code" is vital. Trusting a “black box” system without understanding how the AI has arrived at its decisions is a dangerous game.

AI tools can handle tasks like scheduling and basic market data analysis, but they’re not suited for complex, high-stakes financial decision-making. When AI makes a mistake on a client file, the burden should undoubtedly fall on the human.