The EU AI Act and a small business. What already applies when your office uses ChatGPT

Most small firms do not build AI, they use it. Which AI Act rules already apply to such a firm, what was postponed in 2026, and what to go through with a lawyer.

In this article
  1. What your firm is under the AI Act
  2. What has applied since 2 February 2025
  3. Transparency since 2 August 2026
  4. What was postponed in 2026
  5. National rules and what to do now

In short

A small business that only uses ChatGPT or Claude is usually a deployer under the EU AI Act, not a provider. Since 2 February 2025 the prohibited practices and the duty to support AI literacy apply, and since 2 August 2026 part of the transparency rules. The rules for Annex III high-risk systems were moved to 2 December 2027, and the details of your own case are worth checking with a lawyer.

The accountant asks whether the AI Act means someone has to sign something, because everyone in the office writes emails with ChatGPT. Nobody knows if it is media noise or a real duty. This article sorts out what the EU AI Act changes for a small business that uses AI but does not build it.

This is not legal advice. We describe dates and duties as they stand in the regulation. What they mean for your firm is worth checking with a lawyer.

What your firm is under the AI Act

The AI Act is Regulation (EU) 2024/1689. Among other roles, it separates providers, who build an AI system and place it on the market, from deployers, who use one in their business.

An accounting office that writes emails in ChatGPT, a salon that answers clients with Claude's help, a shop that generates product descriptions. All of them are usually deployers. Most duties in the regulation fall on providers, the companies that make these tools. The regulation does not cover people who use AI purely privately, outside a business.

What has applied since 2 February 2025

Since that date the general provisions and the prohibited practices of Article 5 apply. It is a closed list. It includes manipulative techniques that can cause someone serious harm, exploiting people's vulnerabilities due to age, disability or economic situation, and scoring people for their social behaviour. It also includes emotion recognition in the workplace and in education, except for medical or safety reasons.

That last point is easier to miss than it seems. Software that judges employees' mood in meetings from a camera or voice can be tempting for a small firm. Before you buy anything like it, ask a lawyer.

Article 4 on AI literacy has applied from the same date. In July 2026 it was amended by Regulation (EU) 2026/1744, known as the Digital Omnibus on AI. Under the new wording, providers and deployers take measures to support the development of AI literacy of their staff and of other people who operate AI systems on their behalf. The article also says plainly that it does not require them to guarantee any specific level of AI literacy of any individual.

For a small business this means in practice that people who use AI at work should know how to use it and what to avoid. How to document that in your case is worth agreeing with a lawyer.

Transparency since 2 August 2026

Article 50 is about transparency. Some duties sit with the provider. A chatbot that talks to people has to be designed so a person knows they are dealing with AI, unless that is obvious.

Some sit with the deployer, which is your firm. If you publish an image, video or audio generated by AI that could pass as real, you have to disclose that it was generated or manipulated by AI. The same applies to text published to inform the public on matters of public interest. There is an exception when the text has gone through human review or editorial control and someone holds editorial responsibility for the publication.

An ordinary product description or an email to a client is a different situation from a news article. Where exactly the line runs in your case, check with a lawyer.

What was postponed in 2026

The Digital Omnibus of July 2026 moved the deadlines for high-risk systems. Systems listed in Annex III, which include tools for recruitment and for assessing employees, fall under those rules from 2 December 2027. Systems built into products covered by EU product safety law follow from 2 August 2028.

If you are planning AI that filters CVs or rates staff, this is that category. A deployer's duties here are clearly bigger than with an ordinary chat. Go through them with a lawyer before you buy, not after.

National rules and what to do now

Each member state sets up its own supervision. In Poland, for example, the Act on artificial intelligence systems was published on 27 July 2026 and set up a market supervisory body for AI. If you sell into other EU countries, your lawyer will know which national authority applies.

Three simple steps are enough to start. List the AI tools anyone in the firm uses, including on private accounts. Write short rules about what never goes into them and who checks the output. Talk to your people once about how to use AI and where it gets things wrong. Where to start with client data is in Client data in ChatGPT.

If you want to go through this together, start with a free AI audit. After a conversation of about 30 minutes you get a written list of 3–5 places where AI can save you time, with the tool for each and what to watch out for with client data. The legal assessment we leave to a lawyer.

What would you like to change in your business?

Leave your number. We will talk about your process and see whether we can help. The first call is free.

When is the best time to call?
Controller: Oleksandr Bondarenko (BondSoft)Privacy PolicyTerms of Service
Legal details

The data controller is Oleksandr Bondarenko, trading as Oleksandr Bondarenko Software Solutions (BondSoft), ul. Jana Heweliusza 11, 80-890 Gdańsk, Poland, NIP 5833576264, e-mail: office@bondsoft.pl.

We process your data solely to contact you about your enquiry and to present an offer — on the basis of your consent (Art. 6(1)(a) GDPR and Art. 398 of the Polish Electronic Communications Law) and our legitimate interest in handling the enquiry (Art. 6(1)(f) GDPR).

Providing the data is voluntary but necessary for us to reply. You may withdraw your consent at any time by writing to office@bondsoft.pl — this does not affect the lawfulness of processing before the withdrawal.

You have the right to access your data, rectify it, erase it, restrict its processing, transfer it, and lodge a complaint with the President of the Polish Personal Data Protection Office.

Full information: Privacy Policy.

Send us an email

The service this text is about

Automation and AI / OCR

Automation of document re-typing, OCR and checking rules, and your data stays with you.