The EU AI Act doesn't have special rules for AI agents. Agents are AI systems, so they fall under the Act's general structure: banned practices, transparency duties, obligations on the general-purpose models underneath, and high-risk rules where an agent is used in a listed sensitive area. The part that matters most for everyday users is Article 50, which since 2 August 2026 requires that people know when they are dealing with an AI and that AI-generated content is marked. The Digital Omnibus delayed the high-risk rules into 2027 and 2028. If you use an agent for your own purposes, the Act places no duties on you.

When did each part of the AI Act start applying?

  1. Prohibited practices and AI literacy duties apply
  2. Governance and general-purpose AI model rules apply
  3. Digital Omnibus (Regulation (EU) 2026/1744) enters into force
  4. Most of the Act applies, including Article 50 transparency
  5. Grace period ends for marking output of generative systems already on the market
  6. Annex III high-risk obligations apply (delayed by the Omnibus)
  7. Annex I high-risk obligations apply (delayed by the Omnibus)

The European Commission's regulatory framework page says most of the Act "became applicable on 2 August 2026, with some exceptions." The Omnibus reached political agreement on 7 May 2026 and entered into force on 27 July 2026. It also adds a ban on AI that generates nudified images or child sexual abuse material from December 2026. We did not fetch the Omnibus text from EUR-Lex directly; the dates above come from the Commission, the implementation timeline and law-firm analyses by Goodwin and Gibson Dunn.

What does Article 50 require?

Article 50 sets four transparency duties:

Paragraph Duty What it means for agent users
50(1) People must be told they are interacting with an AI system, unless obvious An agent that calls or messages a business on your behalf should be identifiable as AI
50(2) Synthetic audio, image, video and text must be machine-readable and marked as artificial AI-generated output carries a marker detectable by software
50(3) People must be told when emotion recognition or biometric categorisation is used on them Applies to systems that analyse faces or voices for these purposes
50(4) Deepfakes and AI text published on matters of public interest must be disclosed, unless a human edited the text Relevant if an agent publishes posts for you

Generative systems already on the market before 2 August 2026 have until 2 December 2026 to meet the 50(2) marking duty.

Does the AI Act apply to me as a user?

Not when you use an agent privately. Article 2(10) says the Regulation "does not apply to obligations of deployers who are natural persons using AI systems in the course of a purely personal non-professional activity."

Two exceptions to keep in mind:

Could an agent be "high-risk"?

Rarely, for personal use. An agent becomes high-risk only if it is used for something listed in the Act, such as credit scoring, hiring or access to essential services (Annex III), or built into a regulated product (Annex I). A personal agent booking restaurants isn't high-risk. An agent a bank uses to decide on loans could be. Those duties now start on 2 December 2027 and 2 August 2028 respectively, after the Omnibus delay.

Why are some agents missing in the EU?

dots on EU ChatGPT Business and Enterprise
28 Oct 2026
Gemini Spark availability
EEA excluded
Apple's stated reason for withholding Siri AI
DMA

Availability decisions aren't only about the AI Act:

  • OpenAI dots reached the EU on Business and Enterprise plans on 28 October 2026 (OpenAI community post), but ChatGPT Pro users in the EEA, UK and Switzerland can't use dots yet.
  • Gemini Spark excludes the EEA, UK, Switzerland and Nigeria.
  • Siri AI isn't available in the EU, and Apple has pointed to the Digital Markets Act, a separate competition law.
  • Meta Muse has no announced EU launch as of November 2026.

Getting around restrictions with a VPN can breach terms of service; see our sister site's country availability tracker for legitimate options.

How does this compare with the US and UK?

The EU has binding transparency duties in force; the US relies on general consumer law (see US AI agent rules); and UK regulators say existing law already applies to agents. For UK readers, Personal Agent UK explains what the EU AI Act means for UK users. Security expectations across all three are summarised in our government guidance overview and the OWASP agentic Top 10 explainer.

Bottom line

For EU users, the AI Act mostly works in the background: providers must tell people when they are dealing with an AI and mark synthetic content, and you have no duties for personal use. The bigger practical protections for your data still come from GDPR, and from the safety settings you choose in the agent itself.