FCA authorised · FRN 724952 0117 325 0027Goods in transit: buy onlineQuote & buy →
Apex Insurance Brokers
Speak to a brokerGet a quote →

Aipedia: Data protection consultants

Advising on agentic AI and data protection

Clients deploying AI agents that act autonomously remain responsible for the personal data those agents use, and the ICO has said autonomy is no excuse for poor compliance. Advice in this area is being given ahead of final guidance, which raises the stakes for consultants.

In short

AI agents can read inboxes, update records, contact customers and use other systems with little human input. On 8 October 2026 the ICO opened a six-week call for evidence on agentic AI, closing 20 November 2026, covering security, transparency, accountability, automated decision-making, fairness and lawful data use. It said the fact AI agents act with autonomy is not an excuse for poor compliance, and it has made enquiries about agentic testing in which some agents reportedly bypassed protections. Final agentic AI guidance is planned for Spring 2027. Consultants advising now need to apply existing law carefully and record the uncertainty.

Using AI in your work? Make sure your professional indemnity covers it. A specialist broker can check your wording and find cover that does.
Get a second opinion on your PIDoes PI cover AI mistakes?

What clients are deploying

Agentic AI goes beyond chatbots that answer questions. An agent is given a goal and tools, and it decides the steps: searching a CRM, drafting and sending emails, booking appointments, issuing refunds, or pulling data from one system into another. Clients may ask whether they can let an agent handle customer complaints end to end, triage job applications, or manage supplier onboarding. Each involves personal data and often decisions about people. Some clients are trialling agents built on general-purpose models; others buy them inside existing software, where the agentic features may be switched on by default.

The data protection questions are familiar, but harder to answer: who is the controller for each step, what personal data the agent can reach, whether its actions amount to solely automated significant decisions, and how a person can find out what the agent did and why.

What the ICO has said

Applying existing law

Until guidance arrives, advice rests on existing duties. The controller must be able to show a lawful basis for each processing purpose, minimise what the agent can access, secure the systems it can reach and explain its processing. If the agent makes significant decisions with no meaningful human involvement, Article 22A applies and the Article 22C safeguards are needed; see advising on automated decisions. A DPIA is likely to be required. Agents that call external services may create new processor relationships and transfers.

What can go wrong

Illustrative scenario: a consultant advises a client that an AI agent can handle customer refund requests, on the basis that staff can review its actions. The agent is given broad CRM access and, when a customer asks about an order, it includes details of another customer's purchase in the reply. Reviews were happening only weekly and in summary form. The client treats the incident as a breach, and then questions why the consultant's advice did not address access controls or the review design.

How PI usually responds

Advice on an agentic deployment is professional advice like any other, and PI usually responds to a negligence claim, subject to the wording. The novelty of the technology does not change the test, but it makes the record of what was known at the time more important. Liability questions around agents are still developing; see agentic AI liability. Consultants who also build or configure agents for clients should check whether that work is within their professional business description and read AI insurance in the UK.

Practical controls

How Apex helps

Apex arranges PI for data protection consultants. A broker can review how your policy treats advice on emerging AI, including agents; see the data protection consultants' AI hub.

Sources

Frequently asked

What did the ICO say about agentic AI in October 2026?

On 8 October 2026 it opened a six-week call for evidence closing 20 November 2026, said that autonomy is not an excuse for poor compliance, and confirmed enquiries into agentic testing. Final guidance is planned for Spring 2027.

Who is responsible when an AI agent mishandles personal data?

The organisation deploying the agent remains responsible as controller for its processing. Vendors may share responsibility depending on their role, but autonomy does not shift accountability away from the deployer.

Are an agent's actions automated decisions?

Some will be. If an agent takes a decision with a legal or similarly significant effect and there is no meaningful human involvement, Article 22A applies and the Article 22C safeguards are required.

Check your PI covers how you use AI

Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.

Start a PI quote Request a callback

Apex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.