Businesses using AI to communicate with customers are on notice. From August 2026, every company deploying chatbots, virtual assistants, or AI-powered customer service tools in the European Union must make one thing crystal clear: you’re talking to a machine, not a human.
Article 50 of the EU AI Act mandates that providers ensure AI systems intended to interact directly with natural persons are designed to inform those individuals that they are engaging with an AI system. The catch? This requirement is waived only if it is evident to a reasonably well-informed, observant, and cautious person, considering the circumstances and context of use.
Translation: if there’s any chance someone might think they’re chatting with a human, you need to spell it out. That innocent greeting—”Hi, I’m Mia, what can I help you with?”—needs to become “Hi, I’m Mia, an AI assistant. What can I help you with?”
The Price of Silence
The stakes are higher than most businesses realise. Non-compliance with transparency obligations for providers and deployers pursuant to Article 50 can result in administrative fines of up to €15,000,000 or, if the offender is an undertaking, up to 3% of its total worldwide annual turnover for the preceding financial year, whichever is higher.
For a company with €500 million in annual revenue, that 3% translates to a potential €15 million penalty. For larger enterprises, the percentage-based calculation means fines could climb even higher. The general principle of the AI Act is that penalties shall be effective, dissuasive, and proportionate to the type of offense.
When “Obvious” Isn’t Obvious Enough
The regulation does include an exemption for situations where AI use is “obvious.” But legal experts warn against assuming your setup qualifies. Providers must ensure that AI systems intended for direct interaction with individuals are designed and developed to inform those individuals that they are engaging with an AI system, and this requirement is waived only if it is evident to a reasonably well-informed, observant, and cautious person.
That’s a deliberately high bar. A chatbot on a website called “AI Helper” might seem obviously automated, but what about a sophisticated voice assistant that sounds eerily human? Or a customer service system that seamlessly handles complex queries? Regulators have made it clear: when in doubt, disclose.
The 2026 Deadline
On August 2, 2026, the EU AI Act will become generally applicable and transparency rules under Article 50 will start to apply. That might sound like a comfortable runway, but businesses need to act now. Implementation isn’t just about updating a few text strings in your chatbot interface.
Companies will need to audit every customer touchpoint where AI plays a role, revise communication templates, update user interfaces, train staff on new disclosure requirements, and document compliance procedures for inevitable audits. EU member states must have also implemented one regulatory sandbox by this point, and national competent authorities will ensure compliance with the transparency requirements.
Beyond Chatbots
Article 50 casts a wider net than customer service bots. Deployers of an emotion recognition system or a biometric categorisation system shall inform the natural persons exposed thereto of the operation of the system. Deployers of an AI system that generates or manipulates image, audio or video content constituting a deep fake shall disclose that the content has been artificially generated or manipulated.
The transparency requirements also extend to AI-generated content. Systems generating deep fakes or text intended to inform the public on matters of public interest must clearly disclose their artificial origin, with limited exceptions for artistic or satirical works.
Why This Matters Beyond Compliance
Strip away the legal jargon, and Article 50 is fundamentally about trust. Consumers have grown increasingly wary of AI interactions, particularly as systems become sophisticated enough to mimic human conversation convincingly. The regulation acknowledges a simple truth: people have a right to know when they’re being served by an algorithm rather than a person.
Forward-thinking businesses are already reframing this not as a compliance burden, but as an opportunity. Transparent AI disclosure can actually strengthen customer relationships. It sets clear expectations, reduces frustration when AI reaches its limits, and demonstrates respect for customer autonomy.
Preparing for Implementation
With enforcement beginning in under a year, companies should be mapping their AI landscape now. That means identifying every system that interacts with customers or generates content, assessing whether each system’s AI nature is genuinely “obvious,” implementing clear disclosure mechanisms, and establishing monitoring systems to ensure ongoing compliance.
The European Commission’s AI Office will encourage and facilitate the drawing up of codes of practice at Union level to facilitate the effective implementation of the obligations regarding the detection and labelling of artificially generated or manipulated content. These codes of practice should provide additional practical guidance before the August 2026 deadline.
The EU AI Act represents the world’s first comprehensive AI regulation, and other jurisdictions are watching closely. Companies that get transparency right won’t just avoid fines—they’ll position themselves as leaders in the responsible deployment of AI technology. Those that gamble on non-compliance or assume they can figure it out later may find that €15 million lesson comes rather more quickly than they expected.