Responsible use of AI

Position

Our starting point.

Artificial intelligence supports legal work; it does not replace professional judgement or change who is accountable. Work product is reviewed by the professional responsible for the matter before it is provided to the client. Where AI contributes, professional responsibility remains with us.

We use these tools deliberately and under supervision. Where a tool is suitable for a defined task, it can support an efficient review of material and help identify issues for a lawyer's attention.

How we use it

Where the tools earn their place.

We apply them to work where the value lies in coverage and consistency rather than in judgement: reviewing large volumes of contracts against a defined set of questions, checking a document set for internal consistency, comparing versions and mark-ups, and locating provisions across a data room. The output is checked against the source by the lawyer responsible for the work.

We use AI tools only where they meet the standard set out on this page. Every proposed change remains visible and is accepted or rejected by the lawyer responsible for the document. Unless expressly agreed otherwise in writing, consumer AI services are not used for client work.

What a tool has to meet

The standard, rather than the product.

The tools available change quickly, so we set out the standard we apply rather than the names of the tools in use. Before a tool is used on client material, we check that the provider's contractual terms provide that:

• client data is not used to train the provider's models or those of any third party;
• client data is not retained longer than needed to provide the service, and is deleted after the service ends;
• processing takes place within the European Economic Area, or under a transfer mechanism that meets the requirements of Chapter V GDPR;
• data is encrypted in transit and at rest, and access to it is restricted;
• the provider's security controls are supported by independent assurance, such as a SOC 2 Type II report; and
• where the provider processes personal data on our behalf, a data processing agreement that meets Article 28 GDPR applies before the tool is used for client work.

We do not adopt a tool whose terms do not meet these conditions, and we keep the terms of the tools we use under review. They are also the questions we would suggest any client asks of its own providers.

Confidentiality

How we protect client information.

Our duty of confidentiality, and the legal professional privilege that protects it, constrain what we are willing to do with any technology, including artificial intelligence. The conditions set out above give effect to that duty; they are not merely a preference. Our privacy statement explains how we handle personal data more generally.

Where a client would prefer no artificial intelligence to be used on a matter, we discuss at the outset how the work can be carried out without it, since that affects how the work is scoped and priced.

Verification

Where the limits are.

Generative systems produce fluent output that is sometimes wrong, and an error is generally presented as confidently as a correct answer. We therefore apply a fixed rule: nothing generated by an AI tool is relied on without verification against the primary source. Statutory references, case law, contractual provisions and factual assertions are checked by a lawyer against the document or the source itself.

We do not use these tools as a substitute for legal advice, to determine a position on a disputed point, or to make judgement calls about a client's interests. Those decisions rest with the lawyer responsible for the matter.

Some illustrations on this website were created with an AI tool. Photographs and portraits were not.

Accountability

Who is responsible.

We remain responsible for the professional assessment and the work product we deliver to the client.

Disclosure

What we tell clients.

Where the use of artificial intelligence is material to how a matter is handled, or where a client asks, we address it directly with the client. We prefer to have that conversation at the start of the matter.

Review

Keeping this current.

We are a deployer, not a provider, of AI systems under the AI Act. We keep the tools we use and the standards applied to them under review, taking account of relevant requirements as they become applicable. We revise this statement where a material change warrants it.