Robo-advisers, AI-driven risk-profiling tools and algorithmic portfolio management are now embedded across the wealth management and retail investment sector. Banks, IFAs and discretionary managers increasingly rely on machine-generated outputs to recommend products and assess a client’s capacity for loss. But when that advice turns out to be wrong, and a client suffers a significant financial loss as a result, the question of who is legally responsible becomes pressing. This article examines how established English professional negligence principles apply to AI-assisted financial advice, the relevant legal and regulatory framework, and how affected investors can pursue a professional negligence claim against the firm behind the advice

What Is AI-Assisted Financial Advice?

AI-assisted financial advice covers a broad spectrum of tools now used by regulated firms: automated “robo-advisers” that generate a portfolio recommendation from an online questionnaire; machine-learning models that flag suitable products from a client’s stated goals and risk tolerance; algorithmic trading and rebalancing systems used within discretionary fund management; and generative AI tools used by human advisers to draft recommendations, suitability reports or research summaries. In each case, a firm authorised by the Financial Conduct Authority (FCA) is offering, or facilitating, a regulated activity, and the fact that a machine rather than a person generated part of the output does not remove the firm’s underlying legal and regulatory obligations.

Does Existing Negligence Law Apply to AI-Generated Advice?

Yes. English law has not created a separate legal category for “AI negligence”. The ordinary principles of professional negligence, developed over decades of case law, apply equally where a firm’s advice process is automated, partially automated, or algorithmically assisted. The starting point remains the House of Lords’ decision in Hedley Byrne & Co Ltd v Heller & Partners Ltd [1964] AC 465, which established that a professional who assumes responsibility for the accuracy of advice, knowing it will be relied upon, owes a duty of care to the recipient. A firm cannot avoid that assumption of responsibility simply because the recommendation was produced or informed by an algorithm rather than typed out by a human adviser. If a client reasonably relied on the output of a regulated firm’s advisory process, and suffered loss as a result of a flawed recommendation, the firm that deployed the technology remains the proper target for a claim, not the software itself.

Establishing a Claim: The Three-Stage Test

As with any professional negligence claim, a case involving AI-assisted advice must satisfy three elements on the balance of probabilities:

• Duty of care: The firm owed the client a duty, arising from the advisory relationship, the regulatory status of the firm, or the terms of engagement.
• Breach: The advice fell below the standard of a reasonably competent firm exercising that function, whether the failure lay in the design, testing, calibration or human oversight of the AI tool, or in the firm’s failure to override an obviously unsuitable output.
• Causation and loss: The breach caused quantifiable financial loss that would not otherwise have occurred.

Crucially, a firm cannot discharge its duty of care merely by pointing to the involvement of a sophisticated system. Courts assess the output and the process against the standard expected of a reasonably competent professional performing that role, in line with the long-standing Bolam-type standard applied throughout professional negligence litigation.

Relevant Legal Principles and Case Law

While no reported English decision has yet addressed liability for a generative AI tool specifically, the courts have already grappled with automated and algorithm-driven investment advice, and the reasoning transfers directly.

O’Hare v Coutts & Co [2016] EWHC 2224 (QB)

In this High Court decision, the claimants argued that Coutts’ discretionary investment recommendations, generated through the bank’s internal risk-rating and suitability process, were unsuitable for their attitude to risk. The court examined in detail how the bank’s systems classified the claimants’ risk appetite and whether the resulting recommendations were suitable. Although the claim ultimately failed on its facts, the case confirms that a firm’s internal risk-assessment methodology, however systematised, is squarely open to judicial scrutiny, and that a mismatch between a client’s true risk profile and an automated or semi-automated output can found a viable cause of action.

Rubenstein v HSBC Bank plc [2012] EWCA Civ 1184

The Court of Appeal held a bank liable in negligence where a financial adviser recommended a product that did not match the client’s stated appetite for capital security, notwithstanding contractual limitation clauses. The case remains the leading authority on causation and remoteness in investment-suitability claims and is regularly applied where an AI-influenced recommendation departs from a client’s declared risk tolerance.

Where a firm exercises discretionary management or advisory discretion, the fiduciary-style standard discussed in Bristol & West Building Society v Mothew [1998] Ch 1 is also instructive: the elevated standard of skill and loyalty owed to a client is not diluted by the use of automated decision-support tools.

The Regulatory Backdrop: FCA Rules and the Consumer Duty

Beyond the common law, FCA-authorised firms deploying AI in advice or portfolio management remain bound by the suitability requirements in COBS 9 and 9A of the FCA Handbook, and, since July 2023, by the Consumer Duty under FCA Policy Statement PS22/9, requiring firms to act to deliver good outcomes for retail clients. A firm that deploys an AI system without adequate testing, human oversight or override mechanisms may find that an unsuitable recommendation amounts to both a regulatory breach and a viable civil claim. FCA findings or Financial Ombudsman Service determinations can provide useful supporting evidence, though they are not a substitute for pursuing compensation through the courts.

Common Scenarios Giving Rise to a Claim

• Risk misclassification: an automated questionnaire or scoring model incorrectly assesses a client as having a higher risk tolerance or capacity for loss than they actually have.
• Overreliance without human review: advisers rubber-stamp an AI-generated recommendation without exercising independent professional judgement, particularly where the output is plainly inconsistent with the client’s circumstances.
• Flawed or poorly trained models: a system trained on unrepresentative data produces systematically unsuitable outputs across a class of clients.
• Failure to disclose limitations: a firm fails to explain that a recommendation was AI-generated or fails to warn of the tool’s known limitations.
• Tax and structuring errors: AI-generated financial planning that feeds into tax-inefficient structures can generate follow-on HMRC exposure, see our sister site on negligent tax advice and HMRC penalties

Limitation Periods for AI-Related Financial Negligence Claims

The ordinary limitation rules apply. Under section 2 of the Limitation Act 1980, a claim must generally be issued within six years of the negligent advice being given. Where a client only later discovers the AI-driven recommendation was flawed, the extended period under section 14A, three years from the date of knowledge, may apply, subject to the 15-year long-stop. Given how technical these disputes can be, early specialist advice is essential to protect your position.

What Compensation Can You Recover?

Damages aim to put the client back into the position they would have been in had the negligent advice not been given, typically the difference between the AI-recommended strategy’s actual outcome and that of a suitable alternative, plus foreseeable consequential losses. Firms offering AI-assisted advice carry the same professional indemnity insurance obligations as any other regulated adviser, so compensation often remains recoverable even where a firm faces financial difficulty. Where a firm has since entered insolvency, our colleagues at windinguppetitionsolicitors.co.uk can advise on a parallel creditor claim.

How LEXLAW Can Help

LEXLAW is a City of London firm of qualified Solicitors and Barristers based in Middle Temple, specialising in high-value contentious financial negligence claims against IFAs, private banks, discretionary managers and other regulated firms. Our team regularly assesses complex investment-suitability disputes, including those involving automated advice tools, and negotiates directly with firms, their compliance teams and their professional indemnity insurers to secure optimal settlements, or litigates before the County Court, High Court and Court of Appeal where necessary.

If you have lost money after relying on AI-assisted or algorithm-driven financial advice, find out how to start a professional negligence claim, read our wider professional negligence FAQs, or check the merits of your case with our specialist team. You can also contact us in London to arrange an initial conference, or request a second opinion on an existing case.