Short answer
For many online stores, the parts of the EU AI Act most often discussed are the transparency duties in Article 50, which have applied since 2 August 2026, and the AI literacy duty in Article 4. Product search, recommendations and customer-service chat are not named in the Act's list of high-risk uses, though classification depends on each system's intended purpose, and the Digital Omnibus (Regulation (EU) 2026/1744) pushed the high-risk rules back to December 2027 and August 2028. The Act has no separate rules for AI agents; the Commission's AI Act Service Desk says, in views it calls preliminary, that the existing rules for AI systems and general-purpose AI models cover them. This is general information, not legal advice.
Key takeaways
- Chatbot and AI-content transparency rules apply from 2 August 2026.
- Typical shopping and recommendation uses are not named on the high-risk list.
- The Commission's service desk says existing AI system and model rules cover AI agents.
What is the EU AI Act?
The EU AI Act is Regulation (EU) 2024/1689, the European Union's law on artificial intelligence. It sorts AI by use and puts most duties on providers, the businesses that develop AI systems or offer them under their own name. It entered into force on 1 August 2024 and applies in stages.
This article is general information, not legal advice. It describes what the published texts say, not what applies to your business. For that, talk to a lawyer.
In broad terms, the Act has these layers:
- Prohibited practices (Article 5). A short list of banned uses, such as harmful manipulation.
- High-risk AI systems (Article 6, Annexes I and III). AI in regulated products or in listed areas such as recruitment and credit scoring. These carry the heaviest duties.
- Transparency duties (Article 50). Rules for AI that talks to people or generates content.
- General-purpose AI models (Chapter V). Duties for providers of large models, with extra rules for models with "systemic risk".
- AI literacy (Article 4). A duty for providers and deployers generally.
When do the EU AI Act rules apply after the Digital Omnibus?
The dates changed in 2026. The Digital Omnibus on AI, Regulation (EU) 2026/1744, was published in the Official Journal on 24 July 2026 and, under its Article 4, entered into force on the third day after publication.
The table shows the application dates in the Act as amended:
| Date | What applies |
|---|---|
| 2 February 2025 | Definitions, AI literacy and prohibited practices (Chapters I and II) |
| 2 August 2025 | General-purpose AI model rules, governance and penalties |
| 2 August 2026 | General application, including Article 50 transparency duties |
| 2 December 2026 | New Article 5 bans on AI systems used or offered to generate non-consensual intimate imagery or child sexual abuse material; deadline for generative AI already on the market to meet Article 50(2) marking |
| 2 December 2027 | High-risk rules for Annex III systems (was 2 August 2026) |
| 2 August 2028 | High-risk rules for Annex I product systems (was 2 August 2027) |
The Omnibus links the high-risk delay to "the delayed availability of standards, common specifications, and alternative guidance". It also rewrote Article 4 and extended some SME relief on fines to small mid-cap companies.
Is your store a provider or a deployer?
Under the Act's definitions, a store using another company's AI tool is typically a deployer of it, not the provider. That matters because most duties fall on providers.
Article 3 defines the two roles:
- A provider is a person or body "that develops an AI system or a general-purpose AI model or that has an AI system or a general-purpose AI model developed and places it on the market or puts the AI system into service under its own name or trademark".
- A deployer is a person or body "using an AI system under its authority except where the AI system is used in the course of a personal non-professional activity".
If you build your own assistant, even on top of another company's model, and run it under your own name, you may be its provider. Article 25 also lists cases where a deployer becomes the provider of a high-risk system, for example by putting its own name or trademark on one.
Which EU AI Act rules are most likely to reach an online store?
For a typical store: Article 50, Article 4, and whether anything you use is high-risk.
Telling people they are talking to AI. Article 50(1) says: "Providers shall ensure that AI systems intended to interact directly with natural persons are designed and developed in such a way that the natural persons concerned are informed that they are interacting with an AI system, unless this is obvious from the point of view of a natural person who is reasonably well-informed, observant and circumspect, taking into account the circumstances and the context of use". This duty sits with providers.
Marking and labelling AI content. Article 50(2) requires providers of generative AI systems to ensure outputs "are marked in a machine-readable format and detectable as artificially generated or manipulated". Article 50(4) puts duties on deployers. Deployers of a system that creates a deep fake "shall disclose that the content has been artificially generated or manipulated". Article 3(60) limits deep fakes to image, audio or video content that resembles real persons, objects, places or events and would falsely appear authentic. The text duty in Article 50(4) is narrower: it covers text "published with the purpose of informing the public on matters of public interest", and does not apply where the text has had human review or editorial control and someone holds editorial responsibility.
Article 50(5) says this information must be given "at the latest at the time of the first interaction or exposure".
Prohibited practices. Article 5 bans the use of AI for listed practices, not only its supply, including purposefully manipulative or deceptive techniques that distort behaviour and cause, or are reasonably likely to cause, significant harm.
AI literacy. Since the Omnibus, Article 4(1) says: "Providers and deployers of AI systems shall take measures to support the development of AI literacy of their staff and other persons dealing with the operation and use of AI systems on their behalf". It adds: "This obligation does not require providers or deployers to guarantee any specific level of AI literacy of any individual." The original wording asked for "a sufficient level of AI literacy".
High-risk uses. Annex III lists the high-risk areas: biometrics, critical infrastructure, education, employment, essential private and public services, law enforcement, migration and border control, and justice and democratic processes. It does not name product search, recommendations or customer-service chat, but classification turns on each system's intended purpose. A store could still touch the list, for example through emotion recognition, AI "intended to be used for the recruitment or selection of natural persons" and AI "intended to be used to evaluate the creditworthiness of natural persons or establish their credit score" (fraud detection is excepted). Article 6(1) separately covers AI that is a safety component of products under the EU laws in Annex I.
How does the EU AI Act treat AI agents that visit and buy from your store?
The Act has no chapter, definition or separate rules for AI agents. Its only mention, added by the Omnibus, is "Agentic AI" as one technology code in Annex XIV, a list that conformity assessment bodies use when applying to be designated.
The Commission's AI Act Service Desk says in its FAQ that "while AI agents are not a separate category of AI under the AI Act, the definitions of an AI system in Article 3(1) AI Act and of a GPAI model in Article 3(63) AI Act are sufficient to cover AI agents. This means that the rules applicable to AI systems and GPAI models under the AI Act also apply to AI agents."
The FAQ names the most relevant rules: the Article 5 bans on harmful manipulation and exploitation of vulnerabilities, the Article 50 transparency rules "if the AI agent is intended to interact with natural persons or generate content", and the high-risk rules if an agent is classed as high-risk. It also says "the European Commission's regulatory considerations are only preliminary at this stage."
So the Act's duties for a shopping agent sit with whoever provides it and any business that deploys it. A consumer using an agent for personal shopping is not a deployer under Article 3(4), and Article 2(10) excludes such purely personal use. The FAQ does not say what duties, if any, fall on a website an agent visits or buys from. For the practical side of agent orders, see rules for AI agent orders and prompt injection and shopping agents.
What should you check?
Questions to work through with your advisers, not a compliance checklist:
- List your AI. Every AI tool on your site and in your back office: chat, product copy, images, search, fraud screening, hiring.
- Find your role for each one. Are you using a vendor's system (deployer) or offering one under your own name (provider)?
- Check customer-facing chat. Ask your vendor how its system tells people they are talking to AI, as Article 50(1) requires of providers.
- Check generated media. Where do you publish AI-generated images, audio or video that could look real, and how is it disclosed?
- Check Annex III. Confirm whether any tool you use falls in a listed area, such as recruitment or credit scoring.
- Plan AI literacy. How do staff who run AI tools learn what those tools do and where they fail?
- Track the dates. High-risk rules now start on 2 December 2027 and 2 August 2028.
For a short summary, see the EU AI Act glossary entry. On who agents are and who they act for, see Know Your Agent.
How Sill helps
Sill does not make any business compliant with the EU AI Act, and it does not decide which rules apply to you. It shows you which AI agents and bots visit your site, the pages they read, and what they ask for and buy. When an AI agent places an order, your rules approve it, decline it or hold it for a person to review, and every decision is signed and recorded. Signed audit bundles can be exported and checked by anyone using Sill's published public key.
Questions people ask
Does the EU AI Act apply to a store based outside the EU?
It can. Article 2 says the Act applies to providers and deployers located outside the EU "where the output produced by the AI system is used in the Union". Whether that covers your setup is a question for a qualified adviser.
Did the Digital Omnibus delay the chatbot disclosure rules?
No. Article 50 still applies from 2 August 2026. Its one change to Article 50 timing gave providers of generative AI systems placed on the market before that date until 2 December 2026 to meet the Article 50(2) marking duty.
Is a product recommendation engine high-risk under the EU AI Act?
Product recommendations are not named among the use cases in Annex III, which is where the Act lists high-risk uses such as recruitment, credit scoring and biometrics, including emotion recognition. Classification turns on each system's intended purpose, so a tool used in a listed area, or as a safety component of a regulated product, could still be high-risk.
What are the fines under the EU AI Act?
Article 99 sets maximum fines of up to EUR 35 million or 7% of total worldwide annual turnover, whichever is higher, for prohibited practices, and up to EUR 15 million or 3% for breaches of other listed duties, including the Article 50 transparency obligations. These are maximums: Member States set the penalties actually imposed. For SMEs, the lower of the two figures applies, and since the Omnibus the same is true for small mid-cap companies for fines other than those for prohibited practices.
Sources
- EUR-Lex: Regulation (EU) 2024/1689 (Artificial Intelligence Act), Official Journal text · checked 17 Nov 2026
- EUR-Lex: Regulation (EU) 2026/1744 (Digital Omnibus on AI), Official Journal text · checked 17 Nov 2026
- European Commission AI Act Service Desk: How are AI agents addressed within the AI Act? · checked 17 Nov 2026