The UK has no general AI law. A UK business using AI is governed by existing rules, mainly UK GDPR as amended by the Data (Use and Access) Act 2025, consumer and equality law, and sector regulators such as the FCA. The EU AI Act is different: it is a single, detailed law, and it can apply to a UK company that sells AI products into the EU or whose AI output is used there.
For most UK small businesses that simply use tools like ChatGPT or Copilot for internal work, the practical duties are modest. For a UK software company selling to European customers, or a firm using AI to screen job applicants for roles in the EU, the AI Act needs real attention, though the toughest rules were pushed back in 2026.
- The EU AI Act entered into force on 1 August 2024 and most of it has applied since 2 August 2026. It covers providers outside the EU when their AI is placed on the EU market or its output is used in the EU.
- A 2026 amending law, the AI Omnibus, moved the main high-risk deadlines to 2 December 2027 and 2 August 2028.
- Transparency duties, such as telling people they are talking to an AI, have applied since 2 August 2026.
- In the UK, the King’s Speech of 13 May 2026 again included no standalone AI bill. The ICO is preparing a statutory code of practice on AI and automated decision-making.
Source: European Commission, AI Act policy page, updated 19 October 2026
This is a general guide, not legal advice. If you are a provider of an AI system used in the EU, or you use AI for decisions about people, take specialist advice on your own facts.
How the UK regulates AI
The UK has chosen to regulate AI through existing laws and regulators rather than a single act. The approach set out in the 2023 AI regulation white paper asks regulators to apply five principles within their own sectors: safety, security and robustness; transparency and explainability; fairness; accountability and governance; and contestability and redress.
Labour’s 2024 manifesto promised binding regulation for the companies building the most powerful models. That has not happened. IAPP reported that the King’s Speech of 13 May 2026 “notably omitted a standalone AI bill”. Instead, the government announced a Regulating for Growth Bill which, according to GOV.UK, will introduce statutory sandboxing powers so businesses can test new products, including AI products, in controlled real-world conditions.
The bodies a small business is most likely to meet are:
- The Information Commissioner’s Office (ICO). In practice, the UK’s main AI regulator for anything involving personal data.
- Sector regulators. The FCA for financial services, the CQC and MHRA for health, the SRA for law firms, Ofcom for online services.
- The Competition and Markets Authority and Trading Standards. For misleading claims, including claims about what your AI does.
- The AI Security Institute. Renamed from the AI Safety Institute on 14 February 2025, it tests advanced models for national security risks. It doesn’t regulate businesses.
UK data protection: the rules that bite now
If your AI handles personal data, UK GDPR applies in full, and the Data (Use and Access) Act 2025 changed it. The ICO says all of the Act’s data protection provisions are now in force, phased in between June 2025 and June 2026. Two changes matter most for AI:
- Automated decisions. Significant decisions made solely by automated means are now allowed on more lawful bases, so long as you apply safeguards, including telling people and letting them contest the decision and reach a human. The wider permission does not extend to special category data such as health information.
- Complaints. You must make it easy to complain about how you use personal data and acknowledge complaints within 30 days.
The ICO consulted on updated guidance on automated decision-making from 31 March to 29 May 2026. In October 2026 it said it had secured data protection changes from major AI developers, including Amazon, Anthropic, Apple, DeepSeek, Google, Meta, Microsoft and OpenAI, and opened a call for evidence on agentic AI that closed on 20 November 2026. Both feed into a forthcoming statutory code of practice on AI and automated decision-making. For the practical checks, see our AI agents and UK GDPR checklist.
What the EU AI Act is
The AI Act (Regulation (EU) 2024/1689) sorts AI uses into four tiers, according to the European Commission:
- Unacceptable risk: banned. The Commission lists practices such as social scoring and untargeted scraping of facial images to build recognition databases. Emotion recognition in workplaces and schools is also banned, with narrow exceptions.
- High risk: strict obligations. Uses that can seriously affect health, safety or fundamental rights, including employment, education, access to essential services and critical infrastructure. Providers need risk management, data quality controls, logging, documentation and human oversight.
- Transparency risk: disclosure duties. Telling people when they are interacting with AI, and labelling deepfakes and certain AI-generated text published on matters of public interest.
- Minimal risk: no specific rules. The Commission gives spam filters and AI in video games as examples. Most business use of AI sits here.
Separate rules apply to providers of general-purpose AI models, the large models from companies like OpenAI, Google and Anthropic. Those duties fall on the model makers, not on businesses that use their products.
The timeline after the 2026 AI Omnibus
In November 2025 the Commission proposed an “AI Omnibus” to simplify and delay parts of the Act. It was published in the Official Journal as Regulation (EU) 2026/1744 and, according to the Commission, entered into force on 27 July 2026, days before the main rules were due to start.
| Date | What applies |
|---|---|
| 1 August 2024 | AI Act enters into force |
| 2 February 2025 | Bans on prohibited practices; AI literacy duty |
| 2 August 2025 | Rules for general-purpose AI model providers; governance |
| 27 July 2026 | AI Omnibus in force; AI literacy duty softened |
| 2 August 2026 | Most of the Act applies, including transparency duties |
| 2 December 2026 | New ban on AI that creates non-consensual sexual images of real people; deadline for machine-readable marking by generative AI systems already on the market |
| 2 December 2027 | High-risk rules for standalone uses such as recruitment, education and credit (previously 2 August 2026) |
| 2 August 2028 | High-risk rules for AI built into products covered by EU safety law, such as machinery and medical devices |
According to legal summaries of the amending regulation, the AI literacy duty changed from requiring providers and deployers to “ensure” staff AI literacy to taking measures to support it, without setting any required level. The Omnibus also extended some lighter documentation and quality-system rules to small and medium-sized firms and to a new category of small mid-caps.
Does the EU AI Act apply to a UK business?
It can. Article 2 covers providers that place AI systems on the EU market “irrespective of whether those providers are established or located within the Union or in a third country”, and providers and deployers outside the EU “where the output produced by the AI system is used in the Union”. A provider is whoever develops an AI system and puts it on the market under its own name. A deployer is a business that uses one.
Some typical UK situations:
- You use ChatGPT or Copilot internally to draft documents for UK clients. Very little, if anything, in the AI Act applies to you. UK GDPR still does.
- You run a customer chat agent on a website used by EU customers. The duty to make clear it is an AI falls on the provider of the system, so if you build your own, it falls on you. If you buy one, check the supplier handles it, and say so plainly anyway.
- You publish AI-generated images or video of real-looking people or events to EU audiences. As a deployer, you must disclose that deepfake content is artificial.
- You sell a SaaS product with AI features to EU customers. You are a provider. Check whether any feature falls into a high-risk category, and meet the transparency duties now.
- You use AI to screen candidates or monitor staff in EU countries. Employment is a high-risk area. From 2 December 2027, deployers will need human oversight, logs and to inform affected people. Providers outside the EU of high-risk systems must appoint an authorised representative in the EU.
If you are building an AI product for European buyers, these questions will come up in sales due diligence well before any regulator asks. Our guide to security and compliance for a UK SaaS covers what buyers ask for.
Fines
Article 99 sets maximum fines as a fixed sum or a share of worldwide annual turnover. For most companies the higher of the two applies. For small and medium-sized enterprises, including start-ups, the lower applies.
Enforcement against small firms is likely to start with the clear-cut cases: banned practices and missing disclosures. In the UK, the ICO can already fine for data protection breaches involving AI, which for most small firms is the more immediate risk.
A practical compliance check
Step 1 of 6
List tools
Write down every AI tool in use, who uses it, and what data goes in.
Step 2 of 6
Map the EU
Note any tool whose output reaches EU customers, staff or the public.
Step 3 of 6
Classify
Mark each use as banned, high risk, transparency or minimal, using the Commission's examples.
Step 4 of 6
Disclose
Tell people when they talk to an AI and label synthetic images, audio and video.
Step 5 of 6
Check data
Run a DPIA for personal data and add safeguards for automated decisions.
Step 6 of 6
Review yearly
Revisit the list each year and when you add a tool or enter a new market.
Most of this belongs in your AI policy. If you use AI to process enquiries or customer data, the questions in our guide to AI chatbots for small business are a good starting list. And if you market services that use AI, keep your claims accurate; our article on what AI agents are covers the “agent washing” trap.
Common questions
Is the UK going to copy the EU AI Act?
Not on current plans. The government has preferred regulation through existing regulators and, as of the May 2026 King’s Speech, had not brought forward a general AI bill. UK firms that trade with the EU may still find it simplest to meet the EU standard.
Does using ChatGPT make us a provider?
No. Using a general-purpose tool for your own work makes you, at most, a deployer. You could become a provider if you build an AI system on top of a model and offer it to others under your own name, or substantially change a high-risk system.
Do we need an AI literacy programme?
Under the amended EU rule, providers and deployers must take measures to support AI literacy, without a fixed standard. Sensible training on what your tools can and can’t do, and what data must never go into them, is good practice in the UK too.
Will the deadlines move again?
They might. The high-risk dates have already moved once. Check the Commission’s AI Act page before planning around a specific date.
Where to start
Make the list of AI tools and uses this month, mark anything that touches EU customers or staff, and fix the disclosures first. If you are building an AI product and want a clear view of your obligations before you raise money or sign European customers, our startup consulting team can help you plan it, and our R&D tax relief primer explains how the build costs may qualify.
Sources
- European Commission: AI Act, regulatory framework for AI (updated 19 October 2026, checked 22 November 2026)
- EU AI Act, Article 99: Penalties (text of Regulation (EU) 2024/1689, checked 22 November 2026)
- EU AI Act, Article 2: Scope, and Article 50: Transparency obligations (Regulation (EU) 2024/1689)
- Regulation (EU) 2026/1744 (Digital Omnibus on AI), Official Journal, 24 July 2026, as summarised by nicfab.eu
- ICO: The Data (Use and Access) Act 2025: what does it mean for organisations? (checked 22 November 2026)
- ICO: ICO secures changes from leading AI developers as scrutiny extends to AI agents (8 October 2026)
- IAPP: King’s Speech signals diffuse UK digital policy agenda, but no AI bill (14 May 2026)
- GOV.UK: UK to become world’s fastest market to commercialise innovation with regulatory shake-up (8 July 2026)
- GOV.UK: Tackling AI security risks to unleash growth and deliver Plan for Change (14 February 2025)