Guide · Regulation
What the ICO expects when your team uses AI
DPIAs, records of processing, contracts, staff monitoring and breaches: the UK GDPR duties that apply when a small firm uses AI tools, explained simply.
By Stuart Kinsey, Co-founder of SimpleKPI4 min read

If your team uses AI tools with client data, UK GDPR already sets out what you need to do. There's no special AI law to wait for. This is what the regulator expects, duty by duty, and how a small firm can meet each one without a compliance department.
First, a change of name
On 30 September 2026, the Information Commissioner was replaced by the Information Commission, under the Data (Use and Access) Act 2025. It still goes by the ICO, now short for the Information Commission's Office, and its guidance carries on. ICO announcement
1. Do a DPIA before you rely on AI
A data protection impact assessment looks at the risks of a kind of processing and how you'll reduce them. The ICO's guidance on AI is direct: "In the vast majority of cases, the use of AI will involve a type of processing likely to result in a high risk". High-risk processing needs a DPIA. If you decide your use isn't high risk, write down why.
A good AI DPIA covers which tools you use, what data goes into them, which plans and contracts you're on, and the protections in place.
ICO: guidance on AI and data protection
2. Add AI providers to your records of processing
Article 30 of UK GDPR asks you to keep records of what personal data you process, why, and who receives it. There is an exemption for organisations with fewer than 250 employees, but it falls away if the processing isn't occasional, if it's likely to put people at risk, or if it includes special category or criminal offence data. Client work is rarely occasional, so most firms still need records, and AI providers belong in them as recipients.
3. Have a lawful basis, written down
For most firms the lawful basis for using AI on client work is legitimate interests. That means doing a legitimate interests assessment: what you're trying to achieve, whether AI is necessary for it, and whether your interest is balanced against your clients' rights.
4. Use contracts, and mind transfers abroad
An AI provider handling personal data for you should be under a contract with a data processing agreement, which is what business plans provide. Most AI providers process data in the US, so you also need safeguards for the transfer and a transfer risk assessment explaining why it's acceptable. Free and personal accounts have none of this. We compare the plans in free vs business AI plans.
5. Tell people
Your privacy notice should say that you use AI providers to help with your work, and list them as recipients. Clients shouldn't learn about it for the first time from someone else.
6. Be fair when you check how staff use AI
Checking how staff use AI tools is monitoring, and the ICO allows it when it's necessary and proportionate. It expects you to tell staff before it starts, explain what is recorded and why, and cover it in your DPIA. Recording what kind of data was involved, rather than what people typed, is far less intrusive.
7. Record every breach, and report the serious ones
If personal data ends up somewhere it shouldn't, such as a free AI account, that's a personal data breach. Every breach must be recorded internally. If it's likely to put people's rights and freedoms at risk, report it to the ICO without undue delay and, where you can, within 72 hours of finding out. If the risk to people is high, tell them too.
Has the ICO acted on this yet?
As of 21 October 2026, we're not aware of any ICO enforcement action over staff putting personal data into AI tools. That isn't a reason to wait. Every duty above applies now, and they're the questions a regulator, an insurer or a client will ask after something goes wrong.
A short checklist
- A DPIA for your AI use, reviewed when you add a tool
- AI providers in your records of processing and privacy notice
- A legitimate interests assessment and a transfer risk assessment
- Business plans with a data processing agreement
- An AI policy staff have acknowledged, and a staff privacy notice
- A breach register, including near misses
Saiela's compliance centre drafts these from how your team actually uses AI and flags them when something changes. The AI policy and breach register are on every plan. The DPIA, records of processing and the evidence pack that exports them together are on Team. Whether you're an accountancy practice, a law firm or an estate agency, the duties are the same. See the compliance centre.
General information, not legal advice. For advice on your situation, speak to your data protection lead or a solicitor.

