The EU AI Act is the world’s first comprehensive law for artificial intelligence. It sets rules for how AI systems are built, sold and used in the European Union — and it affects companies far beyond Europe. If your business uses chatbots, AI-generated content, hiring software or any AI tool that reaches people in the EU, this law matters to you.
In 2026 the rules moved from paper to practice. Enforcement powers and transparency duties started on 2 August 2026, while a major amendment called the “AI Omnibus” gave companies extra time for high-risk systems. This simple guide explains what the law is, who it covers, the risk categories, the updated timeline, the fines and seven practical steps to get ready.
⚖️ Key Takeaways
- The EU AI Act (Regulation (EU) 2024/1689) entered into force on 1 August 2024 and applies in stages.
- It uses a risk-based approach: banned AI, high-risk AI, transparency-risk AI and minimal-risk AI.
- Bans on unacceptable AI practices have applied since February 2025; transparency rules since 2 August 2026.
- The AI Omnibus delayed high-risk rules to 2 December 2027 (stand-alone) and 2 August 2028 (AI inside regulated products).
- Fines can reach €35 million or 7% of worldwide annual turnover for the most serious breaches.
What Is the EU AI Act?
The EU AI Act is a European Union regulation that creates one set of rules for artificial intelligence across all 27 member states. Because it is a regulation rather than a directive, it applies directly in every EU country without each one having to write its own version.
The main goal is to make sure AI used in Europe is safe, transparent and respects fundamental rights, while still allowing innovation. Instead of regulating every AI tool the same way, the law looks at how risky a use is. A spam filter faces almost no rules, while AI that decides who gets a job or a loan faces strict requirements.
You can think of the EU AI Act as doing for AI what the GDPR did for personal data: setting a global benchmark that many companies follow everywhere because it is easier than running different systems for different markets. If you are new to data rules, see our guide to GDPR compliance for websites.
EU AI Act at a Glance
Who Does the EU AI Act Apply To?
The law covers several roles along the AI supply chain:
- Providers: companies that develop an AI system or model and place it on the EU market under their own name — whether they are based in Europe or not.
- Deployers: businesses and organisations that use AI systems in their work, such as a bank using AI to screen loan applications.
- Importers and distributors: companies that bring AI products into the EU or sell them there.
- Product manufacturers: firms that build AI into products such as machines, medical devices or toys.
Like the GDPR, the EU AI Act has extraterritorial reach. A company in the United States, Pakistan or India can be covered if its AI system is used in the EU or its AI output is used there. That is why global technology firms such as Microsoft have built dedicated compliance programmes.
Some uses fall outside the law, including AI used only for military or national security purposes, AI used purely for scientific research and development, and personal, non-professional use by individuals.
The 4 Risk Levels Explained
The heart of the EU AI Act is its risk pyramid. The higher the risk, the stricter the rules.
| Risk level | What it means | Examples |
|---|---|---|
| Unacceptable risk | Banned completely | Social scoring, manipulative AI, most untargeted facial image scraping |
| High risk | Allowed, but with strict requirements | AI in hiring, credit scoring, education, medical devices, critical infrastructure |
| Transparency risk | Must tell people AI is involved | Chatbots, deepfakes, AI-generated text and images |
| Minimal risk | No new obligations | Spam filters, AI in video games, inventory forecasting |
Most AI tools used by everyday businesses fall into the minimal or transparency categories. The heavy compliance work is concentrated on banned and high-risk uses.
Banned AI Practices Under the EU AI Act
Since 2 February 2025, the EU AI Act has prohibited AI practices considered a clear threat to people’s safety and rights. These include:
- AI that uses manipulative or deceptive techniques to distort people’s behaviour and cause significant harm
- AI that exploits vulnerabilities linked to age, disability or social or economic situation
- Social scoring that leads to unfair treatment of people
- Predicting whether a person will commit a crime based only on profiling or personality traits
- Untargeted scraping of facial images from the internet or CCTV to build facial recognition databases
- Emotion recognition in workplaces and schools, except for medical or safety reasons
- Biometric categorisation to infer sensitive traits such as race, political opinions or sexual orientation
- Real-time remote biometric identification in public spaces by police, except in narrow, strictly defined cases
💡 Did you know? The bans were not delayed by the 2026 amendments. Breaking them carries the highest fines in the law.
High-Risk AI Systems: What Changes for Businesses
High-risk AI is allowed, but it must meet strict requirements before and after it reaches the market. There are two main groups:
- Stand-alone high-risk systems (Annex III): AI used in sensitive areas such as recruitment and employee management, education and exams, access to credit and essential services, migration and border control, law enforcement, justice and critical infrastructure.
- AI built into regulated products (Annex I): AI that acts as a safety component in products already covered by EU safety laws, such as medical devices, machinery, lifts, toys and vehicles.
Providers of high-risk AI under the EU AI Act must, among other things:
- Set up a risk management system covering the whole life of the AI
- Use high-quality training data and check it for bias
- Keep detailed technical documentation and automatic logs
- Give clear instructions to deployers
- Design the system for effective human oversight
- Ensure accuracy, robustness and cybersecurity
- Pass a conformity assessment, apply CE marking and register in the EU database
Deployers also have duties: using the system according to instructions, assigning trained staff for human oversight, monitoring performance, keeping logs and, in some cases, carrying out a fundamental rights impact assessment. Strong security matters too, because a compromised AI system can be misused — our guide to ransomware attacks shows how quickly attackers exploit weak systems.
Transparency Rules for Chatbots and AI Content
Article 50 of the EU AI Act sets transparency duties that became applicable on 2 August 2026. They are the rules most likely to affect ordinary websites and marketing teams:
- Chatbots and AI assistants: people must be told they are interacting with AI, unless it is obvious.
- AI-generated content: providers of generative AI must mark synthetic audio, images, video and text in a machine-readable way so it can be detected as AI-made.
- Deepfakes: deployers must clearly disclose that image, audio or video content has been artificially created or manipulated.
- AI-generated public-interest text: text published to inform the public must be labelled as AI-generated unless it has been reviewed by a human who takes editorial responsibility.
- Emotion recognition and biometric categorisation: people exposed to these systems must be informed.
The European Commission published guidelines on these transparency obligations in July 2026 to explain the scope and exceptions. If your business uses tools like ChatGPT, Copilot or image generators, read our explainer on generative AI to understand how they work.
Rules for General-Purpose AI Models
General-purpose AI (GPAI) models are large models that can perform many tasks, such as the models behind popular chatbots. Since 2 August 2025, their providers have had to:
- Prepare technical documentation for regulators and downstream developers
- Put a policy in place to respect EU copyright law
- Publish a summary of the content used to train the model
The most powerful models, classed as having “systemic risk”, face extra duties such as model evaluations, adversarial testing, serious incident reporting and strong cybersecurity. A voluntary GPAI Code of Practice helps providers show compliance. From 2 August 2026, the Commission’s AI Office gained full powers to enforce these rules, including fines. These models run on huge amounts of computing power, much of it supplied by chipmakers such as NVIDIA.
EU AI Act Timeline: Key Dates (Updated for 2026)
The EU AI Act applies step by step. Here is the current timeline after the 2026 amendments:
| Date | What applies |
|---|---|
| 1 August 2024 | Law enters into force |
| 2 February 2025 | Bans on unacceptable-risk AI and AI literacy provisions |
| 2 August 2025 | GPAI model obligations, governance rules and national authorities |
| 2 August 2026 | Article 50 transparency duties; AI Office enforcement powers and penalty framework active |
| 2 December 2027 | Stand-alone high-risk AI systems (Annex III) |
| 2 August 2028 | High-risk AI embedded in regulated products (Annex I) |
What Is the AI Omnibus?
In November 2025, the European Commission proposed a “Digital Omnibus on AI” to simplify parts of the EU AI Act. The main reason was practical: the technical standards and testing bodies needed to certify high-risk systems were not ready in time for the original August 2026 deadline.
After negotiations, the Parliament and Council formally adopted the changes. The amendment, Regulation (EU) 2026/1744, was published in the Official Journal on 24 July 2026 and entered into force on 27 July 2026. Its main effects are:
- A delay for stand-alone high-risk AI rules from 2 August 2026 to 2 December 2027
- A delay for high-risk AI inside regulated products to 2 August 2028
- Formal definitions for small and medium-sized enterprises (SMEs) and small mid-cap companies, with simplified documentation and lighter burdens
- Permission to process personal data where strictly necessary to detect and correct bias in AI systems, with safeguards
Importantly, the Omnibus is not a pause. The bans, GPAI rules and transparency duties continue on schedule.
EU AI Act Fines and Penalties
The EU AI Act sets three main tiers of fines, calculated as a fixed amount or a percentage of global annual turnover, whichever is higher:
| Violation | Maximum fine |
|---|---|
| Using banned AI practices | €35 million or 7% of worldwide turnover |
| Breaking most other obligations (high-risk, transparency, GPAI) | €15 million or 3% of worldwide turnover |
| Giving incorrect or misleading information to authorities | €7.5 million or 1% of worldwide turnover |
For SMEs and start-ups, the lower of the two amounts applies, which keeps penalties more proportionate. National market surveillance authorities enforce most rules, while the Commission’s AI Office handles general-purpose AI models.
7 Essential Steps to Prepare for the EU AI Act
- Make an AI inventory: list every AI system and tool your business builds, buys or uses — including AI features inside software such as CRM, HR and marketing platforms, many of which run in the cloud.
- Classify each system by risk: decide whether it is banned, high-risk, transparency-risk or minimal-risk under the EU AI Act.
- Identify your role: are you a provider, deployer, importer or distributor? Your obligations depend on it.
- Stop anything prohibited: if any use falls into a banned category, end it immediately.
- Fix transparency now: add clear “you are chatting with AI” notices, label deepfakes and AI-generated public content, and check that your AI vendors mark synthetic media.
- Train your people: make sure staff who use or oversee AI understand how it works, its limits and the risks.
- Prepare for high-risk deadlines: use the extra time to build documentation, data governance, human oversight and vendor contracts before December 2027.
✅ Quick tip: the official AI Act Service Desk from the European Commission offers free tools to check which obligations apply to your business.
What the EU AI Act Means for Small Businesses and Bloggers
Most small businesses will not build high-risk AI. For them, the EU AI Act mainly means being honest about AI use. If you run a website with a chatbot, tell visitors it is AI. If you publish AI-generated images that look real, label them. If you use AI to write news-style articles, have a human review and take responsibility for them.
Small companies also benefit from support measures: priority access to regulatory sandboxes (controlled testing environments), simplified documentation, lower fees for conformity assessments and proportionate fines.
EU AI Act vs GDPR: How They Work Together
The two laws overlap but do different jobs. The GDPR protects personal data, while the EU AI Act regulates AI systems based on risk. An AI hiring tool, for example, may need to comply with both: the GDPR for processing candidates’ data and the AI Act for being a high-risk system. Businesses that already have strong GDPR processes — data maps, impact assessments and clear privacy notices — have a head start.
How the EU AI Act Affects Companies Outside Europe
The law reaches far beyond the EU’s borders. A software company in Asia that sells an AI recruitment tool to a European employer counts as a provider. An agency in the Middle East that runs a chatbot for EU customers may need to meet the transparency rules. Providers outside the EU that offer high-risk systems or general-purpose AI models must usually appoint an authorised representative inside the EU.
This “Brussels effect” means many international firms are choosing to build AI that meets European standards by default, because maintaining separate versions for different markets is costly. Other governments, including the UK, Canada, South Korea and several US states, are also developing AI rules, often borrowing ideas such as risk categories and transparency labels.
Benefits and Criticisms of the Law
Supporters say the law builds public trust in AI, protects people from harmful uses, creates one clear rulebook for the whole EU market and encourages responsible innovation. Clear rules can also make it easier for businesses to buy AI tools with confidence.
Critics argue the rules are complex, compliance is expensive, and strict requirements could slow European innovation compared with the US and China. The 2026 delays for high-risk systems were partly a response to these concerns, as standards were not ready. Civil society groups, on the other hand, worry that delays and exceptions weaken protections for people.
EU AI Act FAQs
What is the EU AI Act in simple words?
It is a European law that sorts AI systems by risk. Dangerous uses are banned, risky uses must meet strict rules, some uses must be disclosed, and most everyday AI faces no new rules.
When did the EU AI Act come into force?
It entered into force on 1 August 2024. Its rules apply in phases between February 2025 and August 2028.
Was the EU AI Act delayed?
Only partly. The 2026 AI Omnibus delayed high-risk rules to December 2027 and August 2028, but the bans, general-purpose AI rules and transparency duties were not delayed.
Does the EU AI Act apply to companies outside the EU?
Yes. It applies to providers and deployers outside the EU if their AI systems are placed on the EU market or their output is used in the EU.
Do I need to label AI-generated content?
In many cases, yes. Deepfakes must be disclosed, and AI-generated text published to inform the public must be labelled unless a human has reviewed it and takes editorial responsibility.
Does using ChatGPT at work make my company subject to the EU AI Act?
Using a general-purpose chatbot for everyday tasks is usually minimal risk. But if you use AI for high-risk purposes, such as screening job candidates, or deploy a chatbot that talks to customers, extra obligations can apply.
Who enforces the EU AI Act?
National market surveillance authorities in each EU country enforce most rules. The European Commission’s AI Office supervises general-purpose AI models and coordinates enforcement.
What are the penalties for breaking the EU AI Act?
Fines go up to €35 million or 7% of global annual turnover for banned practices, €15 million or 3% for most other breaches, and €7.5 million or 1% for supplying false information.
Conclusion
The EU AI Act is reshaping how businesses around the world build and use artificial intelligence. Bans are already in force, transparency duties and enforcement powers began in August 2026, and high-risk rules arrive in late 2027 and 2028 after the AI Omnibus. The smartest move is to start now: map your AI tools, classify their risk, add clear disclosures and train your team. Companies that treat compliance as a way to build trust will be best placed to win customers in the age of AI.
📚 Keep learning: Explore more guides in our Tech Regulations section.
Sources: Regulation (EU) 2024/1689 (Artificial Intelligence Act); Regulation (EU) 2026/1744 (AI Omnibus); European Commission AI Act guidance. This article is for general information only and is not legal advice. Consult a qualified lawyer for advice on your specific situation.