Does using AI count as outsourcing under the SRA rules?

There is no settled answer to whether an AI tool is "outsourcing" in the regulatory sense, and for most firms the label is not the useful question. What matters is that the obligations which bite on outsourced work bite here too: the firm remains accountable for the work produced, client information must stay confidential, and someone competent must supervise the result. If you apply those obligations properly, the classification question largely answers itself.

What firms usually mean by outsourcing

In practice, firms use the word for two different things.

The first is sending out part of the legal work itself: instructing a costs draftsman, a paralegal agency, a document review provider, an overseas support team. The work is done by people exercising some judgement, and the firm has to satisfy itself that they are competent, that client information is protected and that the arrangement does not stop the firm meeting its own obligations.

The second is buying a service that supports the work: case management software, a dictation service, a cloud document store, an e-signature platform. Here nobody is exercising legal judgement on your behalf, but client data is still being handled by someone else.

An AI assistant sits closer to the second category than the first. It does not take instructions, form a view and send work out under its own name. It produces material that a fee earner reads, checks and decides what to do with. But it does touch client papers, and it does affect the quality of what leaves the firm, so the diligence you apply should be at the serious end of the software scale rather than the trivial end.

The obligations that apply either way

The Solicitors Regulation Authority expects firms to remain accountable for work carried out on their behalf, to have effective systems and controls, to supervise work competently, and to keep client affairs confidential. None of that changes depending on whether a tool is labelled an outsourcing arrangement.

So the practical tests are:

• Competence. The person signing the work must be able to tell whether it is right. If nobody in the firm can check the output, the firm should not be producing it.
• Supervision. Someone with the right level of experience reviews what the tool produced before it is used or sent.
• Confidentiality. Client information must not be exposed to people outside the firm who have no business seeing it, and must not be reused for anyone else's benefit.
• Accountability. The firm, not the supplier, answers to the client, the court and the regulator for the work.
• Records. You should be able to show, after the event, how the work was produced and who checked it.

If you can satisfy all five, the regulatory position is sound whichever label you choose. If you cannot, calling it "just software" will not help you.

The due diligence to do before you sign

Treat the supplier the way you would treat any third party that handles client files.

1. Who is the contracting entity, and where is it? A named company, registered somewhere you can enforce against.
2. Where is the data held and processed? Ask for countries, not marketing phrases. If personal data is involved, UK GDPR requires a written contract setting out what the processor may and may not do.
3. Is your firm's information kept separate from other customers' information? Ask how.
4. Is anything used to improve the product for other users? The answer you want is no.
5. Who inside the supplier can see your material, and under what controls?
6. What happens to the data if you leave, or if the supplier does? Export, deletion, notice periods.
7. Can the firm see what was asked and what was answered? You need this for file records and for any later complaint or claim.
8. What does the contract say about liability and about changes to the service?

Keep the answers. A short supplier file, dated and signed off by whoever owns the decision, is usually enough to show you took the question seriously.

Where the analogy with outsourcing breaks down

There is one important difference, and it cuts against complacency rather than in favour of it. When you instruct an agency, you are relying on a professional who carries their own duties and, usually, their own insurance. When you use an AI tool, no second professional stands behind the work. The whole of the responsibility sits with the fee earner who uses it.

That has a practical consequence for how you supervise. With an agent, you review the finished product and ask questions. With AI, you need to check the underlying material: the page a point came from, the section of the legislation relied on, the arithmetic behind a date. A tool that shows you those things is far easier to supervise than one that simply asserts an answer, and far easier to defend if the work is ever questioned.

Where Alesis fits

Alesis is an AI assistant for UK law firms, made by Alesis.ai Limited and used through the web browser, with one conversation for a matter. It answers questions about a matter from the matter's own papers and names the page each answer came from, and when it cannot find support for a point it says what is missing instead of guessing, which is what makes supervision practical. A firm's information is held in the UK and processed only in the UK and the EU, kept apart from every other firm, and never used to train anything for anyone else. It assists qualified professionals rather than replacing them, prepares drafts for a qualified person to review and sign off, and does not provide legal advice.