The short version: On 10 June 2026, the European Commission published a Code of Practice on marking and labelling AI-generated content. This is part of the EU AI Act's implementation framework and covers any content generated by AI that interacts with the public — text, images, audio, and video. The UK is not inside the EU AI Act's primary jurisdiction, but UK businesses that export to EU customers, operate in EU markets, or use AI-generated content in materials visible to European audiences need to understand the requirements.
What the Code of Practice covers
The core requirement is disclosure: if content has been substantially generated by an AI system, people who encounter it must be able to identify it as such. This applies to text, synthetic images, AI-generated audio (including voice cloning), and video content. The Code provides technical standards for implementation — including metadata watermarking standards — and guidance for businesses on what constitutes adequate disclosure.
The Code of Practice is initially a voluntary instrument — companies operating in the EU AI ecosystem are invited to sign it and implement it. However, it is designed as the precursor to mandatory requirements under Articles 50 and 52 of the EU AI Act, which are scheduled to take effect across EU member states progressively through 2026 and 2027.
Why this matters even if you are in the UK
Brexit means the EU AI Act does not directly apply to UK-based businesses operating solely in the UK market. But several categories of UK business are affected:
- Exporters and e-commerce businesses that sell goods or services to EU customers are already subject to EU AI Act provisions where AI is used in the customer-facing transaction process
- Businesses with EU subsidiaries or staff operating across the UK-EU border
- Businesses that use EU platforms (website hosting, payment processing, marketing platforms) where the platform operator's compliance obligations flow through to business customers
- Businesses that aspire to EU market access — establishing compliant processes now is cheaper than retrofitting later
The UK government's current approach under its own AI strategy is principle-led rather than rule-based — meaning UK businesses do not face mandatory AI content labelling domestically yet. But the EU rules set the de facto standard for businesses that operate across the Channel, and the UK regulatory direction has historically converged with EU standards over time in areas affecting consumer-facing digital activity.
What types of AI-generated content are affected
The most commonly discussed examples are deepfake images, synthetic voice audio, and AI-written articles published without human authorship disclosure. But the scope is broader: marketing images generated by AI tools, product descriptions generated by AI, AI-drafted customer emails sent on behalf of a company, and social media content created by AI systems are all within the category of "AI-generated content interacting with the public."
For UK small businesses, the most relevant practical question is: do you use AI to create any content that a customer or potential customer will see — and do you currently disclose this? If the answer to the first part is yes and the second part is no, the Code of Practice is a prompt to start thinking about how you would handle disclosure.
What disclosure means in practice
The Code does not require prominent disclaimers on every piece of AI-assisted content. The standard is proportionate to the level of AI involvement and the context:
- AI-assisted content (a human wrote it, AI helped edit it) — disclosure is encouraged but the level of requirement is lighter
- AI-generated content (AI wrote it, a human reviewed it) — clear marking is required under the Code
- Fully automated AI content (no meaningful human review before publication) — the strongest disclosure requirements apply
The watermarking standard included in the Code allows metadata to carry AI provenance information invisibly — the content itself does not require a visible label in all cases, as long as the technical standard is implemented.
What UK small businesses should do
Your AI content compliance checklist
Audit your AI content use: List where you use AI to create customer-facing content — website copy, social media, email, product descriptions, images, audio.
Check your EU customer exposure: Do you have EU customers, or could you? If yes, the EU AI Act's content provisions are relevant to you.
Develop a simple disclosure policy: Even a sentence in your website footer noting that some content is AI-assisted demonstrates good faith and is ahead of where most small businesses currently are.
Follow the UK parallel: The UK's own AI content guidance is expected to evolve over the next 12 months. The EU Code of Practice is the likely reference point for UK regulators, so understanding it now positions you well.
No urgent action for UK-only businesses: If you serve UK customers only and have no EU market exposure, this Code does not create immediate obligations. It is a planning signal for 2027 compliance readiness.
The broader context
The EU AI Act content labelling Code of Practice is one of four implementation instruments the European Commission is publishing in 2026. The others cover high-risk AI systems, general-purpose AI models, and interaction with existing data protection law. Together, they represent the most systematic regulatory framework for AI in any major market.
For UK businesses, the message is practical rather than alarming: you do not need to comply today, but you should understand what the EU is requiring and start building the habits that will make compliance straightforward when the rules formalise. The businesses that treat AI content disclosure as a trust signal now — rather than a compliance burden — will be better positioned regardless of how UK-specific rules develop.