What Regulated Professionals Need to Know About Using AI
Regulated professionals can use AI to support summarisation, document preparation, research planning and administrative checking — but the professional remains responsible for confidentiality, accuracy, legal and factual verification, and anything submitted in their name. AI must not file, send or make rights-affecting decisions without authorised human approval. This is general information, not legal or regulatory advice.
Regulated professionals — solicitors, accountants, financial advisers, surveyors, insurance brokers, HR consultants and others — are being sold AI tools that promise to save hours of administrative effort. Some of those promises are genuine. But the sales pitch rarely addresses the part that matters most: who is responsible when the AI is wrong?
Professional regulation exists because clients, courts, markets and the public trust the person who has accepted the duty, not the software they used. AI does not hold a practising certificate. It does not carry professional indemnity insurance. It cannot appear before a court, tribunal or regulator to explain an error. That responsibility stays with the qualified professional — regardless of how the work was produced.
AI may assist professional work. It does not inherit the professional's duty, judgement or accountability.
The Quick Answer
Regulated professionals can safely use AI for many supporting tasks. The key rules are:
- Sensitive client information should enter only approved, assessed systems — not free consumer AI tools
- AI-generated legal, factual or source citations must be verified independently before any use
- Deadlines, court dates and case references must be checked against primary sources, never trusted from AI output
- AI must not file, send correspondence or take any rights-affecting action without explicit authorised human approval
- Professional responsibility for accuracy and client interests cannot be delegated to an AI system
Last checked: 6 August 2026. Professional regulatory requirements change; verify the current rules for your specific profession and sector.
Where AI genuinely helps in regulated work
The question is not whether regulated professionals should use AI — many already do, and the efficiency gains for certain tasks are real. The question is which tasks are safe to delegate, which require close supervision, and which must stay entirely with a qualified person.
| Task type | AI role | Human responsibility |
|---|---|---|
| Summarising long documents | Can produce a first summary for the professional to review | Check completeness, accuracy and any legal nuance before relying on it |
| Drafting standard correspondence | Can produce a first draft from a template or instructions | Read, verify facts, check tone and suitability for the specific client |
| Research planning | Can suggest research directions or sources to check | Verify every source independently; AI cannot reliably cite case law or statute |
| Administrative checking | Can flag missing fields, formatting issues or obvious gaps | Sign off every check; AI can miss context-specific requirements |
| Advice, opinions or recommendations | Must not be delegated — these require qualified professional judgement | The professional alone owns the advice |
Confidentiality and data handling
Most professional regulatory frameworks impose strict duties of confidentiality that apply to client information regardless of how it is processed. Entering a client's personal details, financial information, medical history or legal circumstances into a free consumer AI tool is almost certainly a breach of those duties, and may also engage data protection obligations under UK GDPR.
Before using any AI tool with client information, assess it using the same due diligence you would apply to any supplier who will hold your clients' data: who is the legal entity, where is data processed and stored, can model training be turned off, what do the terms say about data use, and is the tool on your firm's approved list?
A free AI tool that is acceptable for general business writing is not automatically suitable for information covered by professional confidentiality duties.
The hallucination risk is higher for professional work
AI language models generate text that is plausible-sounding, not necessarily accurate. For general business writing, a factual error is a nuisance. For regulated professional work, it can have serious consequences: a cited case that does not exist, a statute quoted with the wrong section number, a deadline stated incorrectly, a regulatory requirement that has been repealed.
The professional obligation to verify does not change because AI produced the draft. Every factual assertion, legal reference, source citation and key date in AI-assisted work must be checked against a primary source before it is relied on. If it cannot be verified, it cannot be used.
Sector-specific frameworks
The guide in our AI Governance hub uses solicitors in England and Wales as the clearest worked example, because their regulatory framework is well documented — covering duties to clients, courts and third parties, and the SRA's regulatory expectations. Similar principles apply across regulated professions, each with their own sector-specific rules:
- Solicitors — SRA Code of Conduct, duties to the court, client care obligations
- Accountants — ICAEW, ACCA and other body standards; tax advice obligations
- Financial advisers — FCA requirements, suitability obligations, MiFID principles
- Surveyors — RICS professional and ethical standards
- HR professionals — CIPD guidance alongside employment law obligations
- Insurance brokers — FCA conduct requirements and fair customer outcomes
This guide does not apply any profession's rules identically to another. If you are uncertain how AI use interacts with your specific regulatory obligations, check with your professional body or take appropriate advice.
Read the full guide
Our AI Governance hub contains the full detailed guide to using AI safely in regulated professions, including the worked examples for solicitors, the confidentiality framework, and the controls that keep professional responsibility where it belongs:
Using AI Safely in Regulated Professions — full guide →
AI Governance Knowledge Centre — policies, guides and templates →
Plain-English Takeaway
AI can support regulated professional work — but it cannot accept a professional duty, carry indemnity insurance or explain itself to a regulator. Sensitive information should enter only approved systems, critical sources and deadlines must be checked independently, and AI must not file, send or make rights-affecting decisions without authorised human approval. This is general information, not legal or regulatory advice.
Related Articles
Building Safe AI Workflows in Professional Services
Law firms, accountancy practices, consultancies and other professional-services businesses spend significant time on intake, document preparation, records entry and administrative checking. AI can reduce that cost — but only when the workflow is designed carefully, the data is controlled, and a qualified professional retains responsibility for every important decision. This article covers how to identify the right tasks, what controls a safe workflow requires, and where the human review point must sit.
Read articleAI Recruitment Tools: What the ICO Found When It Looked
In November 2024 the Information Commissioner's Office published the results of a set of consensual audits into the providers and developers of AI tools used for sourcing, screening and scoring job candidates. It made almost 300 recommendations, and the audited organisations accepted or partially accepted every one of them. The findings read like a ready-made due-diligence list: tools that let recruiters filter out people with protected characteristics, tools that guessed gender and ethnicity from a name, tools that hoovered up far more personal data than they needed and kept it indefinitely. For a UK SME thinking about buying one of these tools, that report is a free head start — the regulator has already told you where the problems tend to hide. This article summarises what was found, where employment law meets training data, and the questions worth putting to any vendor before you sign.
Read articleThe Small Business That Automated Its Biggest Mistake
Artificial intelligence rarely invents a brand-new risk. More often it takes a weakness the business already had — a habit, a shortcut, an unchecked step — and runs it faster and further than a person ever could. A quoting error that once reached one customer can now reach two hundred in an afternoon, each message personalised, polished and convincing. This article follows a fictional but realistic example to show how the same underlying mistake becomes far more dangerous once it is automated, and it sets out three ordinary controls that would have caught it before it spread.
Read article