What Is the EU AI Act?

The EU AI Act (Regulation (EU) 2024/1689) is the European Union’s law on artificial intelligence. It sorts AI systems by the risk they pose, bans a short list of practices outright, sets strict duties for high-risk systems, adds obligations for providers of general-purpose AI models and requires transparency when people deal with AI or AI-generated content. It applies to anyone who places AI on the EU market or whose AI output is used in the EU, wherever the company is based.

This entry is a plain-language summary for orientation, not legal advice. Dates and obligations below were checked against the European Commission’s pages in November 2026. For a specific system, read the regulation text and take advice from a qualified lawyer.

How the Act Classifies AI

CategoryExamplesWhat applies
Prohibited practicesSocial scoring, manipulative techniques that cause harm, certain biometric uses, and since the 2026 amendments, apps that generate non-consensual intimate images or child sexual abuse materialBanned
High-riskAI used in hiring, education, credit scoring, critical infrastructure, border control; AI that is a safety component of regulated productsRisk management, data governance, documentation, logging, human oversight, accuracy and security requirements, conformity assessment
Transparency riskChatbots, voice agents, systems that generate text, images, audio or video, deepfakesDisclosure and marking duties (Article 50)
Minimal riskSpam filters, most internal productivity toolsNo specific duties beyond AI literacy

General-purpose AI models, the foundation models behind most GenAI products, have their own chapter: technical documentation, information for downstream providers, a copyright policy and a training-content summary, with extra duties for models that pose systemic risk.

Roles: Provider and Deployer

The provider develops an AI system or model and places it on the market under its own name. The deployer uses an AI system in its own business. A company that builds a chatbot on a hosted model and offers it to customers is usually the provider of that chatbot; the company that uses it internally is a deployer. Both roles have duties, and an organisation can hold both.

Timeline, as Amended

The Act entered into force on 1 August 2024 and applies in stages. In 2026 it was amended by the “AI Omnibus” (proposed by the Commission on 19 November 2025, politically agreed on 7 May 2026, in force from 27 July 2026), which moved the high-risk deadlines back. The current dates, per the Commission’s AI Act page and Article 113 as amended:

DateWhat applies
2 February 2025Prohibited practices; AI literacy duty for providers and deployers
2 August 2025Obligations for general-purpose AI models; governance and penalty rules
2 August 2026General application, including Article 50 transparency duties; national authorities and the AI Office supervise and enforce
2 December 2026New prohibition on AI that generates non-consensual sexually explicit content or child sexual abuse material; deadline for generative systems already on the market before 2 August 2026 to apply machine-readable marking (Article 111)
2 August 2027Deadline for general-purpose AI models placed on the market before 2 August 2025
2 December 2027High-risk rules for Annex III uses (biometrics, critical infrastructure, education, employment, migration and border control and others)
2 August 2028High-risk rules for AI embedded in products covered by EU product safety law (Annex I, for example machinery, toys, lifts)

Earlier articles that cite 2 August 2026 as the high-risk deadline are out of date. Checked in November 2026.

Transparency Duties Most Products Meet First

Under Article 50, already in application:

  • A system that talks with people, such as a chatbot or voice agent, must make clear they are dealing with AI, unless that is obvious.
  • Providers of systems that generate synthetic audio, images, video or text must mark the output in a machine-readable way so it can be detected as AI-generated.
  • Deployers must disclose deepfakes, and AI-generated text published to inform the public on matters of public interest, unless a person with editorial responsibility has reviewed it.

Penalties

Under Article 99, fines reach up to EUR 35 million or 7% of worldwide annual turnover for prohibited practices, up to EUR 15 million or 3% for most other breaches, and up to EUR 7.5 million or 1% for supplying incorrect or misleading information to authorities, whichever is higher. For SMEs and start-ups, the lower of the two amounts applies.

How It Relates to Other Rules

The AI Act does not replace the GDPR. Personal data used by an AI system is still governed by data protection law, including rules on international transfers. The Act is one input to a company’s wider AI governance.

In Practice

Most business GenAI projects start with three questions: does any use case fall under Annex III, is the client a provider or a deployer, and which Article 50 disclosures and markings apply. The answers become features of the system: logs that keep the records a high-risk system must keep, a review step where a person can override the output, and notices that tell users they are dealing with AI. The final legal classification is made by the client’s counsel, and the engineering team supplies the documentation they need to make it.

We help map AI use cases to these obligations in AI strategy consulting, and build the controls into systems through AI security solutions development. Compliance questions also come up in technical due diligence of AI companies.

Back to Glossary

Let's talk about your project

We'll help you harness AI and ML technologies to create innovative software solutions

File requirements: pdf, doc, docx, rtf, ppt, pptx — max 15 MB

No file selected