EU AI Act guide for businesses: what applies and when

A plain-English overview of the EU AI Act, its risk categories, key dates including recent changes, and what small and mid-sized businesses outside and inside the EU should do.

By the UK Technology Hub teamPublished 4 minute read

The short answer

The EU AI Act regulates AI systems by risk. Some practices are banned, high-risk systems face strict requirements, certain systems such as chatbots and AI-generated content carry transparency duties, and general-purpose AI models have their own rules. It can apply to businesses outside the EU whose AI is used in the EU. Key duties include AI literacy since February 2025 and transparency since August 2026, with high-risk deadlines moved to late 2027 and 2028.

The EU AI Act is the first comprehensive AI law from a major economy. Even businesses outside the EU may need to comply if their AI systems are placed on the EU market or their outputs are used in the EU. This overview is general information, not legal advice; check the official texts and take advice for your situation.

Who it applies to

The Act defines roles, including:

  • Providers: develop AI systems or models and place them on the market.
  • Deployers: use AI systems in a professional capacity.
  • Importers and distributors.

Most small businesses using AI tools are deployers; those building AI products may be providers.

Risk categories

Prohibited practices

Certain uses are banned, such as manipulative techniques causing significant harm, social scoring by public authorities and some uses of biometric categorisation and emotion recognition in workplaces and education, with limited exceptions.

High-risk systems

AI used in areas such as employment and recruitment, education, access to essential services, credit scoring, law enforcement and critical infrastructure can be high-risk. Providers face requirements on risk management, data quality, documentation, human oversight, accuracy and robustness; deployers have duties too.

Transparency obligations

  • People must be informed when interacting with an AI system such as a chatbot, unless it's obvious.
  • AI-generated or manipulated content, such as deepfakes, must be disclosed in certain circumstances.
  • Providers of generative AI must mark outputs as AI-generated in machine-readable ways where feasible.

Minimal risk

Most AI uses, such as spam filters or AI-assisted writing, face no specific obligations beyond general laws and AI literacy.

Key dates

  • August 2024: the Act entered into force.
  • February 2025: prohibitions and AI literacy obligations applied.
  • August 2025: obligations for general-purpose AI models applied.
  • August 2026: transparency obligations under Article 50 applied, with a limited grace period for some existing generative AI systems' labelling duties.
  • December 2027 and August 2028: high-risk system obligations, after amendments in 2026 delayed the original dates.

Timelines have changed before, so confirm current dates.

AI literacy

Organisations providing or deploying AI must take measures to ensure staff have sufficient AI literacy. Our AI literacy training guide explains how to meet this practically.

Penalties

Fines can be significant, scaled by the type of breach and company size, with the highest for prohibited practices.

Practical steps for businesses

  1. Inventory AI use: tools, features and custom systems.
  2. Classify risk: prohibited, high-risk, transparency or minimal.
  3. Add transparency: disclose chatbots and label AI content where required. Our guide to AI content labels covers labelling.
  4. Train staff on AI literacy.
  5. Review vendors: ask suppliers about their compliance.
  6. Document decisions and risk assessments.
  7. Monitor updates to guidance and timelines.

Outside the EU

UK businesses serving EU customers, or whose AI outputs are used in the EU, may be affected. The UK and other countries are developing their own approaches.

A worked example

A UK online retailer with EU customers reviewed its AI use against the Act. It used a chatbot on its website, AI-generated product imagery for some lifestyle shots and AI tools internally for drafting copy. The review found that the chatbot needed a clear disclosure that customers were talking to AI, that realistic AI-generated lifestyle images should be labelled, and that staff needed AI literacy training. None of its uses were high-risk. Changes took a few weeks and also improved customer trust.

Deployer checklist

  • List AI systems and their purposes.
  • Check for any prohibited practices.
  • Identify any high-risk uses, such as recruitment or credit decisions.
  • Add AI disclosures for chatbots and other interactive systems.
  • Label deepfakes and certain AI-generated content.
  • Train staff on AI literacy.
  • Keep documentation of assessments and decisions.
  • Ask vendors for compliance information.

Providers have more duties

Businesses that build and sell AI systems, or place general-purpose models on the market, carry more obligations, such as technical documentation, risk management and marking AI outputs. Seek specialist advice if you're a provider.

Relationship with other laws

The AI Act sits alongside data protection, consumer protection, product safety and equality laws. Compliance with the AI Act doesn't replace those obligations.

Staying current

Guidance documents, codes of practice and standards continue to be published, and timelines have already changed once. Assign someone to track updates and review your position at least twice a year.

How we help

Our AI solutions team helps inventory and assess AI use, our AI chatbot team builds assistants with required disclosures and our web app developers add transparency, logging and human oversight to AI features.

Frequently asked questions

Does the EU AI Act apply to UK businesses?

It can, if their AI systems are placed on the EU market or their outputs are used in the EU.

Do I need to label my chatbot as AI?

Under the AI Act's transparency rules, people should be informed they're interacting with AI unless it's obvious.

When do high-risk obligations apply?

After 2026 amendments, high-risk obligations are scheduled for December 2027 and August 2028, depending on the category.

This guide is general information, not legal, tax or financial advice, and details such as prices, features and rules can change. Product and company names mentioned are trademarks of their respective owners and are used only to identify those products; UK Technology Hub is not affiliated with or endorsed by them.

Related guides

All AI & automation guides