From August 2, 2026, the EU Artificial Intelligence Act (AI Act) will impose a clear obligation on companies: the user must know when they are interacting with an algorithm. What does this mean in practice, who does it affect, and how high are the penalties for ignoring the new regulations?
August 2026 is a moment that will permanently change the way companies talk to their customers. This is when key transparency provisions of the EU AI Act will come into effect. No more guesswork – the user must be clearly informed whether there is a human or a bot on the other side of the screen.
What does the AI Act change in AI communication?
The EU Artificial Intelligence Act, adopted in 2024, is gradually organizing the new technology market. For the average internet user, the most important part will be Article 50, which directly regulates the issue of transparency. According to it:
- Companies are required to inform users that a given interaction is taking place with an AI system (unless it is completely obvious from the context).
- This principle applies not only to simple chatbots but also to voice assistants, content generators, or advanced recommendation systems.
- The notification must be simple, clear, and provided directly – hiding it in multi-page terms and conditions will not suffice.
The goal is to rebuild trust online. Especially where our money or health is at stake – that is, in banking, e-commerce, and customer service in general.
Who is affected by the new regulations?
The new law does not only target Silicon Valley giants. This obligation will cover every entity that makes AI systems available to users within the European Union – regardless of where the company is physically headquartered. The following must prepare for the changes:
- Banks and insurers – bots on hotlines or virtual advisors.
- Online stores – intelligent shopping assistants and personalized recommendations.
- Telecoms – automated technical support and ticket handling.
- Media and marketing – text generators or virtual presenters.
- Medical facilities – initial patient triage by algorithms.
Smaller companies will not avoid these obligations either, although EU institutions are announcing support programs and assistance for them in implementing the new procedures.
How should companies inform about the use of AI?
The AI Act does not impose a single, rigid template for the notification. However, the European Data Protection Board (EDPB) points to several best practices in its guidelines:
- In-flight notification: A short message at the very beginning of the chat, e.g., "Hi! You are talking to an AI assistant. How can I help you?".
- Visible labels: Clear labels in the application interface or on the website.
- System notifications: A pop-up notification upon the first launch of a feature based on artificial intelligence.
The most important rule? Language understandable to everyone, without technical jargon and without attempts to hide the truth from the user.
Some brands are already testing such solutions:
- PKO BP: introducing voice messages warning about talking to a bot on the hotline.
- Allegro: testing clear labels in chat windows with assistants.
- Orange Polska: implementing clear notifications in its self-service channels.
Penalties for non-compliance
Ignoring the regulations will be very costly. Financial penalties provided for in the AI Act can amount to:
- Up to 35 million euros or 7% of the company's total global annual turnover – for the most serious safety violations.
- Up to 15 million euros or 3% of global turnover – for lesser infringements, including failure to meet AI-related information obligations.
The stakes are high enough that no one can afford to ignore the new guidelines.
What are the reactions of the industry and consumer organizations?
The tech business declares support for transparency but does not hide concerns about costs and technology:
- Microsoft: announces the gradual implementation of AI labeling standards in its flagship services, such as Copilot.
- Meta: supports a level playing field but points out that smaller companies may have trouble adapting quickly to the new requirements.
- Google: has been developing systems for labeling content generated by algorithms in its services for a long time.
On the other hand, consumer rights organizations view the new law with hope, but also with reservation:
- Panoptykon Foundation: supports the changes while warning against the phenomenon of so-called "AI-washing," i.e., appearing to meet requirements without real transparency.
- Polish Consumers Association: emphasizes that the key to success will be uniform and clear interpretations of the regulations to avoid market chaos.
Practical implementation challenges
Adapting systems to the new requirements is a significant challenge. Companies must face several barriers:
- Costs: According to market analysis, adapting systems to the requirements of the AI Act can cost a medium-sized enterprise from tens to even two hundred thousand euros, depending on the scale of operations.
- Technology: It is necessary to rebuild existing IT systems and integrate new information modules.
Hybridity: It is difficult to precisely mark the moment of transition when a client switches smoothly between a bot and a live consultant during a single conversation.
However, ready-made standards and tools are coming to the rescue:
- Watermarking: Methods of digital content marking, developed, among others, within the C2PA standard.
- AI Detectors: Solutions such as Hugging Face's AI Detector.
- Market standards: Guidelines developed by organizations such as Partnership on AI.
Do similar regulations apply outside the EU?
The European Union is blazing a trail, but the rest of the world is not standing still:
- United States: No single federal law, but individual states are taking matters into their own hands. An example is California and its regulations regarding bot transparency for commercial purposes.
- United Kingdom: Currently focuses on a flexible approach and recommendations, although discussions on tightening the course are ongoing.
- China: Already requires clear labeling of all synthetic content and imposes strict requirements on generative service providers.
Most international organizations, including the OECD and UNESCO, also promote the principles of ethical and transparent artificial intelligence, although their recommendations are mainly advisory.
What does this mean for users?
For all of us, this is a step towards greater safety. We will have the full right to know who we are talking to. This builds awareness and allows us to calmly assess the credibility of the information provided to us. Of course, at the beginning, we may encounter attempts to circumvent the regulations through inconspicuous notifications. In such situations, it is worth reacting and reporting irregularities to the Personal Data Protection Office (UODO).
Summary: what's next?
The entry into force of the regulations in August 2026 is a turning point. Transparency is ceasing to be a good practice and is becoming hard law. Companies that prepare their systems early enough will not only avoid huge fines but will also gain in the eyes of their customers.
If you want to learn more about how technology affects our daily lives, read our article on the impact of artificial intelligence on the brain and on changes in higher education in the era of AI.
Sources
- https://www.polsatnews.pl/wiadomosc/2026-07-13/koniec-z-botami-udajacymi-konsultantow-i-ukrytymi-tresciami-ai-nowe-zasady-od-2-sierpnia/
- https://eur-lex.europa.eu/legal-content/PL/TXT/?uri=CELEX%3A32024R1689
- https://eur-lex.europa.eu/legal-content/PL/TXT/?uri=CELEX%3A32024R1689#d1e3207-1-1
- https://eur-lex.europa.eu/legal-content/PL/TXT/?uri=CELEX%3A32024R1689#d1e3158-1-1
- https://eur-lex.europa.eu/legal-content/PL/TXT/?uri=CELEX%3A32024R1689#d1e3363-1-1
- https://eur-lex.europa.eu/legal-content/PL/TXT/?uri=CELEX%3A32024R1689#d1e3335-1-1
- https://eur-lex.europa.eu/legal-content/PL/TXT/?uri=CELEX%3A32024R1689#d1e3163-1-1
- https://ec.europa.eu/commission/presscorner/detail/pl/ip_26_2145
- https://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill_id=201720180SB1001
- https://leg.colorado.gov/bills/sb23-169
- https://www.gov.uk/government/publications/ai-regulation-a-pro-innovation-approach
Comments