ICO Agentic AI Call for Evidence: What UK Businesses Using AI Agents Should Do
The ICO opened an agentic AI call for evidence on 8 October 2026, closing 20 November. What it means for UK businesses using AI agents, plus a checklist.
On 8 October 2026 the Information Commissioner's Office (ICO) opened a call for evidence on how UK data protection law applies to agentic AI, closing on 20 November 2026. UK businesses using AI agents should treat it as a signal: map what your agents can access, keep people in charge of important decisions, and record your reasoning.
This is general information about regulator activity, not legal advice.
Key takeaways
- On 8 October 2026 the ICO opened a call for evidence on how data protection law applies to agentic AI, closing on 20 November 2026.
- The same day it said it had secured data protection changes from ten major AI model developers and paused engagement with xAI while it investigates Grok.
- The ICO says the fact AI agents act with autonomy is not an excuse for poor compliance, so responsibility stays with the business deploying them.
- Map what your agents can access, keep people in charge of significant decisions and document your risk assessment now.
What happened
The ICO made two linked announcements on 8 October 2026.
Changes secured from AI developers
In its news release, the ICO said it had secured data protection improvements from ten foundation model developers: Amazon, Anthropic, Apple, Cohere, DeepSeek, Google, Meta, Microsoft, OpenAI and Stability AI. Some changes have been made and others committed to. They cover clearer transparency information, stronger ways for people to exercise their rights and more rigorous assessments of safeguards. The ICO says it is monitoring progress.
The release also says the ICO paused its engagement with X.AI after opening a formal investigation into its Grok AI system, which is ongoing.
A call for evidence on agentic AI
The ICO also opened an agentic AI call for evidence, running from 8 October to 20 November 2026. The survey covers:
| Section | Why it matters to a small business |
|---|---|
| Data security | What an agent can reach, and what happens if it is misused |
| Transparency | Whether customers know an agent is handling their data |
| Accountability | Who in your business owns the agent's actions |
| Automated decision-making | Whether the agent makes decisions with real effects on people |
| Fairness and purpose limitation | Whether data is used only for what it was collected for |
| Lawfulness of processing | Your lawful basis for what the agent does |
The ICO says the evidence will help it add clarity and practical support before finalising its guidance. Its news release links this work to a forthcoming statutory code of practice on AI and automated decision-making.
Richard Nevinson, the ICO's Director of Technology Regulation, said: "the fact AI agents act with autonomy is not an excuse for poor compliance." The ICO also says it has made enquiries about reports that some agents bypassed protections, used unauthorised communication channels and accessed external systems.
Who this affects in the UK
AI agents are software that can take actions, not just answer questions: reading an inbox and drafting replies, updating a CRM, booking appointments or chasing invoices. If you use one, or a tool with an "agent" feature switched on, you are the deployer, and data protection responsibilities sit with you. Our explainer on what AI agents are covers the basics.
The risk is highest for businesses handling sensitive or regulated data:
- Dental practices and healthcare clinics, where health data is special category data.
- Law firms and accounting firms, with client confidentiality and professional rules on top of UK GDPR.
- Recruitment agencies, where agents screening candidates may stray into automated decision-making.
Practical checklist
- List your agents. Include built-in AI features in tools you already use, not just bespoke builds.
- Write down what each agent can access. Mailboxes, calendars, customer records, payment systems. Remove anything it does not need.
- Use separate, limited accounts. Give agents their own credentials with the narrowest permissions, and turn on two-factor authentication for the human admin accounts behind them.
- Keep a person in charge of significant decisions. Anything affecting someone's job application, credit, health or legal position needs human review before it happens.
- Do a data protection impact assessment where processing is likely to be high risk, and keep it with your records.
- Update your privacy notice to explain where AI agents process personal data and how people can query a decision.
- Check supplier terms. Find out where data is processed, whether it is used for training and how you can delete it.
- Log what agents do. Keep enough records to reconstruct an action if a customer complains.
- Consider responding to the call for evidence by 20 November 2026 if you deploy agents and have practical experience to share.
Agents that send email on your behalf also depend on correct authentication, or their messages may be junked. Check your domain with our free SPF checker and DMARC checker.
At Make It Flow we map each workflow before automating it and design agents so people stay in charge of judgement and anything regulated. If you want a second pair of eyes on an agent you already run, talk to us.
FAQ
What is the ICO's agentic AI call for evidence?
It is a survey the ICO opened on 8 October 2026 asking how data protection law should apply to AI systems that act autonomously. It covers security, transparency, accountability, automated decision-making, fairness and lawful processing. It closes on 20 November 2026 and will inform the ICO's final guidance.
Do I have to respond to the call for evidence?
No. Responding is voluntary. It is worth doing if you deploy AI agents and have practical experience of the risks or the safeguards that work. The ICO notes that responses from organisations may be published, in full or as a summary, so write accordingly.
Am I responsible if an AI agent mishandles personal data?
In practice, yes. If your business decides to use the agent and for what purpose, you are likely to be the controller. The ICO has said that agents acting with autonomy is not an excuse for poor compliance. Contracts with suppliers help allocate risk but do not remove your duties.
Does this mean small businesses should stop using AI agents?
No. The ICO's message is about using them responsibly, not avoiding them. Limit what agents can access, be open with customers, keep humans reviewing significant decisions, and document your assessment. Many low-risk admin tasks can be automated safely with these controls.
Which AI companies agreed changes with the ICO?
The ICO named Amazon, Anthropic, Apple, Cohere, DeepSeek, Google, Meta, Microsoft, OpenAI and Stability AI. It said some changes are already made and others are commitments it is monitoring. It separately paused engagement with X.AI while it investigates the Grok AI system.
Sources
More guides for your industry
Free checklist (PDF)
Free dental practices website checklist
30 practical checks for dental practices: website speed and mobile, Google Business Profile and reviews, email security, booking and enquiry handling, and AI search visibility. Each one says why it matters, and many link to a free tool that checks it.
Written by the MakeItFlow team
We build the website audit these guides are based on. Every fact is checked against the primary sources listed above and this guide was last reviewed on 25 October 2026 . How we research and check
Check your own site in 30 seconds
Run a free audit covering speed, SEO, DNS, email authentication and security headers — and get a prioritised fix list.
