This Week in Brief

The conversation this week has shifted from excitement about what AI can do to a more grounded assessment of what it actually does — and where it falls short. New research showed that general-purpose AI tools like ChatGPT give dangerously unreliable answers on financial matters, while a major industry analysis concluded that AI is following the same arc as robo-advice: lots of fear, but ultimately a tool that reshapes the profession rather than replaces it. Meanwhile, regulators signalled they are watching unregulated AI money tools very closely.

Key Developments

When AI Gets It Wrong: ChatGPT Fails Broker Safety Checks Two-Thirds of the Time

FT Adviser reported this week that when ChatGPT was tested on broker “safe call” assessments — checking whether firms are legitimate and authorised — it gave wrong answers two-thirds of the time. This is a significant finding for any IFA tempted to use a general-purpose AI tool to quickly assess counterparties or check whether a firm is properly regulated. The headline is a reminder that the confident, polished tone of AI responses can mask serious inaccuracy. For regulated tasks — anything touching client money, authorisation status, or compliance — purpose-built, FCA-aware tools remain the only safe option.

AI and the Robo-Advice Parallels: Reform, Not Replacement

Professional Adviser published a detailed investigation this week drawing direct parallels between the current AI hype cycle and the robo-advice wave of a decade ago. The conclusion: the same fears surfaced, and the same thing happened — advisers weren’t replaced, the tools reshaped the job. The analysis found that only 7% of advice firms are currently using AI in client-facing conversations, with meeting transcription and report-writing the dominant use cases. Crucially, efficiency gains are going into capacity growth rather than fee reductions — firms are looking after more clients with the same headcount, not cutting what they charge. A notable finding: 55% of advisers currently using AI say they are open to switching provider within the next 12 months, suggesting no one has yet built the tool that locks in the market.

FCA and PRA Hold the Principles-Based Line — But Flag a New Risk

Reporting published this week confirmed that the FCA and Prudential Regulation Authority are maintaining their technology-neutral approach to AI: no AI-specific rules, reliance instead on existing Consumer Duty, SM&CR, and operational resilience frameworks. That consistency will reassure firms that the regulatory goalpost isn’t moving. However, a more significant development was the FCA’s flagging of a new concern: the rapid growth of unregulated AI chatbots providing financial guidance to consumers without any regulatory oversight. The FCA is now actively considering where the boundary lies between information and advice when delivered by AI — and enforcement action in this area looks increasingly likely later this year.

Lloyds Puts AI Financial Assistant Through Its Paces with 7,000 Staff

Lloyds Banking Group moved its AI financial assistant from blueprint to beta testing this week, with 7,000 employees now trialling the tool that the bank plans to roll out to its 21 million retail customers. The assistant is designed to give personalised financial coaching via the mobile app, with the ability to handle spending, savings and investment queries and refer on to human support when needed. Lloyds expects AI to deliver over £100 million in value across the group in 2026. For IFAs, this matters because it sets the tone for what mainstream consumers will start to expect from financial guidance — and where the boundaries of regulated advice will increasingly be tested.

The Paraplanner Shortage Isn’t Going Away

FT Adviser ran a piece with the headline “There are not enough paraplanners to cover the roles” — and separately, analysis from NextWealth found that current AI tools only touch fewer than a quarter of the 41 distinct process steps in a typical advice engagement. The tools that exist do the writing-up well; they don’t capture initial client data, run cashflow forecasts, chase letters of authority, or implement recommendations. The practical implication: AI is reducing the administrative load on paraplanners, but it is not replacing paraplanning, and the underlying resource constraint in advice firms has not gone away.

From the Trade Press

  • “ChatGPT ‘safe calls’ on brokers are wrong two-thirds of the time” — New research testing ChatGPT’s accuracy on financial authorisation checks found it wrong in two-thirds of cases, a warning about relying on general AI for regulated tasks. FT Adviser, 8 April 2026

  • “Mastering AI cyber security risks for advice firms” — Practical guidance on the cybersecurity risks that come with adopting AI tools, including data exposure and third-party dependencies. FT Adviser, 7 April 2026

  • “Are AI tools the new robo advisers?” — A thorough investigation into the state of AI in financial advice, drawing parallels with the robo-advice wave and concluding that AI will reform, not replace, the adviser role. Includes an exclusive list of the top 15 most-viewed AI tool reviews by advisers in Q1 2026. Professional Adviser, 7 April 2026

What to Watch

The FCA’s concern about unregulated AI guidance tools operating outside the regulatory perimeter is worth tracking closely over the coming weeks. As more consumer-facing AI money apps arrive — and as clients start using them — the question of where general guidance ends and regulated advice begins is going to become a live compliance issue for IFAs, not just a policy debate.

Sources: FT Adviser, Professional Adviser, web search. Compiled 10 April 2026. �