Compliance · 8 min read · Updated 25 October 2026

Data Security and AI for Law Firms: What the SRA, ICO and NCSC Say

Confidentiality, Article 32 security, impact assessments, transfers and supplier checks for firms adopting AI.

By Appan Pathmanathan, Chief Operating Officer, Lex Sentinel

A law firm using an artificial intelligence (AI) tool must protect client data under the Solicitors Regulation Authority (SRA) Code and the UK General Data Protection Regulation (UK GDPR). The Information Commissioner's Office (ICO) and the National Cyber Security Centre (NCSC) both say security applies to AI as to other systems. In my view, a firm should check the supplier, assess the risk before use and record the decision.

I am Chief Operating Officer of Lex Sentinel, which builds AI tools for conveyancers, so I have an interest in this topic. This article is general information for firms regulated by the SRA in England and Wales and is not legal advice. It draws on the sources listed below, which I checked on 25 October 2026. Several ICO pages say they are under review, so check the current versions of all of them. Statements marked "in my view" or "my suggestion" are mine.

What security duties does the SRA place on a firm?

A law firm must keep client affairs confidential and have effective systems and controls. Paragraph 6.3 of the SRA's Code of Conduct for Firms requires a firm to keep the affairs of current and former clients confidential unless disclosure is required or permitted by law or the client consents. Paragraph 2.1 requires effective governance structures, arrangements, systems and controls, and paragraph 2.3 says the firm remains accountable where work is carried out through others. Paragraph 2.5 requires it to identify, monitor and manage material risks. I have not read 6.4 and 6.5 as security rules, as they concern disclosure of material information and adverse interests.

The SRA's warning notice on the misuse of AI, published on 17 August 2026, applies those duties to AI. It says client information should be entered into AI systems only where appropriate safeguards are in place, and warns that both paid and free tools may not provide the contractual and technical safeguards needed to protect confidentiality. Lex Sentinel's guide to using public chatbots on client files covers the notice in more detail.

How big is the cyber risk for law firms?

Cybercrime was the most commonly chosen risk to their own firm in the SRA's 2023 survey of solicitors, at 49%. The SRA's report on the risk of the legal profession, published on 13 December 2024, draws on a survey of 515 solicitors carried out between September and December 2023. Respondents could pick up to three risks. Cybercrime was also the leading perceived risk to the legal sector (43%), and lawtech and AI came second for the sector (36%). The survey is two years old, so read it as context.

The SRA's cyber security thematic review, from September 2020, visited 40 firms in 2019. Five of them, 12.5%, held Cyber Essentials Plus, and 75% said they had been directly targeted. The firms were chosen at random from firms that had reported cybercrime and the sample was small, so I treat it as historical context.

What does the ICO say about security and AI?

The ICO says security measures depend on the risk, that there is no one-size-fits-all approach, and that AI's effect on security depends on how a system is built and used. Its guidance on security and data minimisation in AI, dated 26 August 2025 and marked as under review, recommends deleting intermediate files that contain personal data. It also covers supply-chain risk from open-source machine learning frameworks and attacks such as model inversion and membership inference.

The ICO's guide to data security, dated 9 September 2025, says Article 32(1) of the UK General Data Protection Regulation (UK GDPR) requires a level of security appropriate to the risk. It calls Cyber Essentials a good starting point of basic controls and says a firm may need to go beyond them. It says controllers must choose processors that give sufficient guarantees, with the Article 32 measures reflected in the contract. The ICO's page on controllers and processors says a processor that decides the purposes and means of processing has the same liability as a controller.

On data protection impact assessments (DPIAs), the ICO's DPIA page, dated 13 October 2025, lists AI as innovative technology, which calls for an assessment when combined with another high-risk indicator, and recommends doing one if in doubt. The ICO's guide to international transfers, updated on 15 January 2026, says the rules apply to all restricted transfers, even small ones, so in my view a supplier that processes data outside the UK needs a lawful transfer route if the arrangement is a restricted transfer.

Several ICO pages are marked under review because of the Data (Use and Access) Act 2025, while the ICO says all the data protection provisions of the Act are now in force (ICO update of 19 June 2026).

What does the National Cyber Security Centre (NCSC) say?

The NCSC says security applies across the whole life of an AI system. It lists prompt injection and hallucination among the weaknesses. Its page on AI and cyber security, dated 13 February 2024, says the burden of using AI safely should not fall on individual users. Its Cyber Essentials overview lists five controls: firewalls, secure configuration, security update management, user access control and malware protection. The page is general and not written for law firms, and it may have been updated since.

What does the Law Society say?

The Law Society's generative AI guidance, updated in September 2025, says firms should check where data is processed, stored and accessed. Its cyber security guidance for solicitors covers multi-factor authentication, backups and Cyber Essentials, and says some insurers ask about a firm's security and that it may lower professional indemnity insurance premiums. I could not see a date on that page.

What should a firm check before using an AI tool?

In my view, a firm should treat an AI supplier like any other processor and check it before it goes live. The sources do not prescribe a checklist, so these steps are my own suggestions.

Identify who is the controller and who is the processor, and get a written contract that reflects the security measures.
Find out where the data is processed and stored, who can access it, whether it is used to train models and how long it is kept.
Confirm the transfer route for any processing outside the UK.
Complete a data protection impact assessment where the processing is likely to be high risk.
Check the supplier's security certifications and your own, and confirm multi-factor authentication and access controls.
Record the decision, and review it when the tool or the supplier changes.

About the author

Appan Pathmanathan is a solicitor who has worked in UK legal practice since 2001, including litigation involving money laundering and fraud, and conveyancing, and has been managing partner of a law firm. He delivers anti-money laundering and compliance training. Profile: LinkedIn.

Sources

Checked on 25 October 2026.

Frequently Asked Questions

What data security duties apply when a law firm uses AI?

A law firm must keep client affairs confidential under paragraph 6.3 of the SRA Code, maintain effective systems and controls, and apply security appropriate to the risk under Article 32 of the UK GDPR.

Do we need a data protection impact assessment for an AI tool?

Yes, where the processing is likely to result in a high risk to individuals. The ICO treats AI as innovative technology and recommends a DPIA if you are in doubt.

Is Cyber Essentials enough for a law firm using AI?

The ICO calls Cyber Essentials a good starting point of basic controls and says a firm may need to go further. In my view, it does not replace assessing the risks of a specific AI supplier.

Can an AI supplier process client data outside the UK?

A firm needs a lawful transfer route for any restricted transfer. The ICO says the rules apply to all restricted transfers, even small, infrequent ones.

Who is responsible if an AI supplier is breached?

In my view, the firm remains accountable. SRA paragraph 2.3 says a firm stays accountable where work is carried out through others, and the ICO says a controller must choose processors that give sufficient guarantees.