EU AI Act for small businesses using chatbots

For most small businesses, the EU AI Act asks three simple things of a customer-service chatbot: tell people they are talking to AI, keep a human in control and make sure the answers are accurate. Much stricter rules apply only when AI makes important decisions about people, such as hiring or credit. Albania is not yet an EU member, but this law can matter to businesses here more than it might seem. This is general information and does not replace advice from a lawyer.

What the EU AI Act is

In 2024 the European Union adopted the Artificial Intelligence Act, which the European Commission describes as the first-ever comprehensive legal framework on AI worldwide. The regulation sets common rules for developing and using AI systems in the EU. It entered into force in August 2024 and applies in phases: prohibited practices since February 2025, the transparency rules that cover chatbots since August 2026, and the rules for many high-risk systems from December 2027 or August 2028, after a 2026 simplification package moved those deadlines. Dates can still change, so always check the official sources.

The main idea is simple: the greater the risk an AI system poses to people's safety and rights, the stricter the rules. The official summary is on the European Commission's page on the AI Act, and the full text with detailed explanations is on the independent resource artificialintelligenceact.eu.

The four risk levels

The Act sorts AI systems into four main groups:

• Unacceptable risk, meaning prohibited practices. For example, social scoring, manipulative techniques that exploit people's vulnerabilities, or emotion recognition in the workplace and in schools, with a few limited exceptions.
• High risk. Systems that affect important decisions in people's lives, such as selecting job candidates, assessing creditworthiness, education or access to essential services. These carry strict obligations: risk management, documentation, data quality and human oversight.
• Limited risk, with transparency obligations. This is where chatbots and AI-generated content belong: people must know they are talking to a machine or that content was created artificially.
• Minimal risk. Spam filters, games or simple recommendations, with no new specific obligations.

Most customer-service assistants, such as a chatbot that answers questions about products, opening hours and delivery, usually fall into the transparency group. The picture changes if the AI is used for decisions that weigh heavily on people.

Does the EU AI Act apply to a small business in Albania?

There are at least three reasons it matters even if your business is in Tirana, Korçë or Prishtina:

1. Reach beyond the EU's borders. The Act applies to AI systems placed on the EU market wherever the provider is based, and in some cases also when an AI system's output is used inside the EU. If your assistant regularly serves customers in Italy, Greece or Germany, discuss it with a lawyer.
2. European integration. Albania, as a candidate country, is aligning its legislation with EU law, and Kosovo is heading in the same direction. Similar rules can reasonably be expected in local legislation.
3. The market standard. European partners, large platforms and customers increasingly expect AI to be used fairly and under control.

What the EU AI Act means for a customer-service chatbot

For an online store, a restaurant or a beauty salon, the requirements come down to a few practical principles.

Tell customers they are talking to AI

A simple sentence in the first message, the chat window or the business profile is often enough. A salon in Tirana could open with: "Hi, I'm the salon's virtual assistant. If you'd prefer, I can connect you with a colleague." Transparency like this builds trust, not suspicion.

Keep a human in control

An AI assistant needs a clear route to your team, especially for complaints, delicate cases or decisions that affect the customer. The team should be able to step into any conversation.

Don't let the AI make things up

Wrong answers about prices, warranties or return conditions can harm both customers and the business. A good system says "I'm not sure" instead of guessing; the full explanation is in why AI makes up answers.

Be careful with generated content

If you use realistic AI-generated images or videos in marketing, such as photos of people who don't exist, consider labelling them; the Act sets transparency obligations for such content.

Train your team

The Act asks organisations that use AI to make sure their staff have sufficient knowledge of these systems. For a small business, this can be a short meeting on how the assistant works, what it shouldn't be trusted with and when to step in.

The Act doesn't replace the GDPR: when AI processes personal data, the privacy rules apply at the same time. For that side, read our guide to GDPR and AI chatbots for customer service.

A short EU AI Act checklist for small businesses

1. List every AI tool you use: customer chat, image generation, analytics, candidate screening.
2. For each one, ask: does it talk directly to people, and does it make decisions that affect them?
3. Tell customers they are talking to an AI assistant.
4. Make sure a person can take over the conversation at any moment.
5. Ask your vendor how the system works and handles data; see how to choose an AI chatbot platform.
6. Consult a lawyer if you serve customers in the EU or use AI for important decisions, such as hiring.

How AsistentiAI fits these principles

Several of these principles are built into AsistentiAI, the AI agent that answers your customers 24/7 in Albanian and in the language they write in. It never makes up prices, stock or policies; when it isn't sure, it tells the customer honestly and hands the conversation to your team with a short summary, marked "Needs you" in the dashboard. All conversations from Instagram, Messenger, WhatsApp and the website chat are in one place, and your team can take over any of them at any time. When a colleague answers a question the agent didn't know, it becomes knowledge only with the business's approval.

Your business's data stays yours; the full policy is on our privacy page. See the agent at work in the live demo or request a free trial with 50 agent replies and no credit card; our team opens your account within the day.

Frequently asked questions

Does the EU AI Act apply to chatbots?

Yes. Chatbots that talk to people generally fall under the transparency obligations: people should know they are interacting with AI.

Is a customer-service chatbot high-risk under the EU AI Act?

Usually not. A chatbot that answers questions about products, opening hours and delivery typically falls into the limited-risk group. That can change if the AI makes decisions with real consequences for people.

Does the EU AI Act apply to businesses in Albania?

It can, for example when an AI system is offered on the EU market or its output is used in the EU, such as an assistant that regularly serves customers in Italy or Germany. Ask a lawyer about your situation.

When do the EU AI Act rules apply?

According to the European Commission, the Act entered into force in August 2024, prohibited practices apply since February 2025 and the transparency rules for chatbots since August 2026. Rules for many high-risk systems follow in December 2027 and August 2028. Check the Commission's page for current dates.

The EU AI Act is not a reason to fear AI, but a guide to using it honestly. For most businesses here, the key requirements are transparency, human control and accuracy; follow them today and you'll be ready when similar rules reach local legislation.