The Data (Use and Access) Act 2025 (DUAA) is the biggest change to UK data protection law since Brexit, and most of it took effect on 5 February 2026. For AI agents, the key change is to automated decision-making: organisations can now make significant decisions without a human on a wider range of lawful bases, as long as safeguards are in place. You keep the right to be told, to make your case, to ask for a human and to contest the decision. Everything else in UK GDPR still applies to agents.
What is the Data (Use and Access) Act 2025?
The Data (Use and Access) Act 2025 amends UK GDPR, the Data Protection Act 2018 and the Privacy and Electronic Communications Regulations (PECR). It also creates frameworks for smart data schemes (such as extending open banking-style sharing to other sectors) and for digital verification services.
The Act was brought in stages. The main data protection provisions commenced on 5 February 2026 under the Commencement No. 6 Regulations (SI 2026/82), made on 29 January 2026. Law firms including DLA Piper criticised the short notice.
- Data (Use and Access) Act receives Royal Assent
- Commencement No. 6 Regulations made (SI 2026/82)
- Most data protection changes take effect, including new ADM rules
- Duty to handle data protection complaints takes effect
- ICO agentic AI call for evidence closes
What changed for automated decision-making?
This is the part that matters most for AI agents. Section 80 and Schedule 6 of the Act replaced the old Article 22 of UK GDPR with a new framework.
Before 5 February 2026
Under the old Article 22, you had a right not to be subject to a decision based solely on automated processing that had legal or similarly significant effects, unless it was necessary for a contract, authorised by law or based on your explicit consent.
After 5 February 2026
Organisations can now rely on any lawful basis, including legitimate interests, for solely automated significant decisions, as long as safeguards are in place. The stricter limits now apply mainly to significant decisions that use special category data, such as health or ethnicity.
The safeguards you are entitled to for a significant automated decision are:
- Be toldYou must be given information about significant decisions made about you.
- Make your caseYou can make representations about the decision.
- Get a humanYou can ask for human intervention.
- Contest itYou can challenge the decision.
The ICO has signalled that it may focus enforcement on automated systems that lack transparency or meaningful human intervention, according to DLA Piper. In other words, the safeguards have to work in practice, not just on paper.
| Old Article 22 (to 4 Feb 2026) | New framework (from 5 Feb 2026) | |
|---|---|---|
| Lawful basis for significant automated decisions | Contract, law or explicit consent only | Any lawful basis, including legitimate interests |
| Special category data | Very limited | Still tightly restricted |
| Right to human intervention | Yes | Yes |
| Right to contest | Yes | Yes |
| Right to information | Yes | Yes |
Simplified. Not legal advice.
What does this mean for personal AI agents?
Most of what personal agents do is not a "significant decision" about you. Booking a table, drafting an email or reminding you about a bill does not have legal or similarly significant effects. So the ADM rules matter less for everyday agent tasks than for, say, an AI that decides whether you get a loan.
They matter more where an agent:
- applies for credit or insurance on your behalf
- is used by a company to decide refunds, account closures or eligibility
- screens job applications or tenancy applications
In those cases, the company using the agent must give you the safeguards above. Our page on your ICO data rights with agents explains how to use them.
What else changed that affects agent users?
Subject access requests
Organisations now only have to make "reasonable and proportionate" searches when you ask for your data, and can pause the clock if they genuinely need you to clarify. This applies when you ask an agent provider for its records about you.
Recognised legitimate interests
Some purposes, including fraud prevention, network security, safeguarding and emergency response, are now presumed legitimate without a balancing test. Agent providers may rely on this for security monitoring. The processing must still be necessary, and you can still object in some cases.
Complaints to the organisation first
From 19 June 2026, organisations must have a process for handling data protection complaints. Complain to the agent provider first, then to the ICO.
Cookies and PECR fines
Consent rules for some low-risk cookies were relaxed, and PECR fines were raised to UK GDPR levels. PECR also covers unsolicited marketing texts and calls, which matters for agents that send messages. Our sister site covers PECR rules for AI text messages.
Smart data
The Act lets government create smart data schemes beyond banking. Over time, that could let a personal agent switch your energy or broadband provider through secure data sharing, in the way open banking works today. See open banking and AI agents.
Does the DUAA explain why agents skip the UK?
Not obviously. On paper, the DUAA makes the UK more permissive than the EU on automated decision-making, and the UK has no equivalent of the EU AI Act (see does the EU AI Act affect UK users). The European Commission also renewed the UK's data adequacy decision, to 27 December 2031, according to DLA Piper.
Yet Meta Muse, OpenAI dots on personal Pro plans and Claude Duet all exclude the UK alongside the EEA and Switzerland. Our reading, which is an inference rather than a vendor statement, is that this reflects caution and operational simplicity, plus the ICO's open review, rather than a hard legal block. See why AI agents are US-only.
What happens next?
The ICO is consulting on updated ADM guidance and preparing a statutory code of practice on AI and automated decision-making. Its call for evidence on agentic AI closes on 20 November 2026 and will feed that work. Expect the code in 2027.
Bottom line
The Data (Use and Access) Act loosened the UK's rules on automated decisions while keeping the core safeguards. For most personal agent tasks it changes little day to day, because booking and drafting are not significant decisions. Where agents do make decisions that affect your money, job or home, you still have the right to information, a human and a challenge.



