.png){kind=logo}
Why most bots are already non‑compliant
More than 70% of European‑facing chatbots still hide their AI nature behind friendly avatars or vague greetings. The EU AI Act’s Article 50, which took effect on August 2, 2026, says that practice is illegal. If a user could reasonably think they are talking to a person, the bot must say it is an AI from the first interaction.
What Article 50 actually requires
The law states:
Providers shall ensure that AI systems intended to interact directly with natural persons are informed that they are interacting with an AI system, unless this is obvious from the point of view of a reasonably well‑informed, observant and circumspect natural person, taking into account the circumstances and the context of use.
In plain English, you must display a clear, understandable disclosure at the moment the conversation starts. The disclosure cannot be hidden in a privacy policy or appear only after several exchanges.
Who the rule applies to
Two roles are covered:
Providers – anyone who builds or integrates an AI system, even under another brand.
Deployers – anyone who operates the AI for end users, such as a business running a support bot on WordPress, Discord, or Telegram.
In practice, most businesses are deployers, so the compliance burden falls on you, not just the underlying technology vendor.
The "obvious" exception is narrow
Regulators interpret "obvious" very strictly. A chat bubble that says "Ask us anything!" with a human‑like name does not qualify. If there is any doubt that a user might think the bot is human, you must disclose.
What doesn’t work / common pitfalls
Many organizations try to meet the rule with shortcuts that do not hold up under scrutiny:
Placing the disclosure in a footer or a terms‑of‑service page – users never see it before interacting.
Showing the notice only once and then hiding it – the law requires the disclosure to be visible throughout the session.
Using legal jargon or tiny fonts – the message must be readable and understandable to a "reasonably well‑informed" person.
Relying on the "obvious" exception for bots with avatars – regulators have made it clear that visual cues alone are insufficient.
These approaches can still trigger fines of up to €15 million or 3 % of global turnover.
Step‑by‑step compliance checklist
Insert a clear opening line. The first message should read something like "Hello, I’m an AI assistant here to help you." Place it before any user‑generated content.
Make the label persistent. Use a visible badge or prefix (e.g., "[AI]") that stays on the chat window for the duration of the conversation.
Write in plain language. Avoid legal terms; keep the sentence under 20 words and use a font size that matches the rest of the UI.
Implement on every channel. WordPress widgets, Discord bots, and Telegram bots each need their own initial disclosure. Do not rely on a single website notice to cover all platforms.
Test with a "reasonable user". Ask a colleague who is not familiar with the bot to interact. If they hesitate about the bot’s nature, your disclosure is insufficient.
Why grounded AI makes compliance easier
When your bot only answers from a curated knowledge base, there is nothing to hide. Adding a one‑line disclosure does not change the user experience, but it does add a layer of trust. Aidevelopia already grounds its responses in your own documentation, so the compliance work is limited to the UI change described above.
Implementation example for three popular channels
WordPress chat widget
Insert the following HTML snippet at the top of the widget’s response template:
Hi, I’m an AI assistant. How can I help you today?
Ensure the CSS class "ai‑badge" is always visible next to the message bubble.
Discord bot
When the bot receives the first "messageCreate" event, reply with:
@Bot is an AI assistant. Ask me anything about our services.
Keep the prefix "[AI]" on every subsequent reply.
Telegram bot
Use the "sendMessage" method to prepend the same disclosure to the first outbound text. Set "parse_mode" to "HTML" and include a persistent "🤖 AI" icon in the header of each message.
What to do if you’re unsure
Run a quick audit this week:
Open each channel as a new user.
Check whether the first message contains a clear AI label.
Confirm the label remains visible after the first exchange.
If any channel fails, add the disclosure and redeploy. The change can be made in under five minutes for most platforms.
FAQs
Do I need to disclose AI on internal tools?
The rule applies to any system that interacts with a "natural person" in the EU. If the tool is only used by employees and not exposed to external users, you may be exempt, but a risk‑based assessment is recommended.
What if my bot is purely voice‑only?
Audio disclosures are acceptable if they are clear and audible from the start of the conversation. A short spoken sentence – "You are speaking with an AI assistant" – satisfies the requirement.
Can I use a logo instead of text?
A logo alone is insufficient unless it is universally recognized as an AI symbol and is accompanied by plain‑language text. Combining a small robot icon with the sentence "I am an AI" meets the standard.
Will the fine apply if I’m a small startup?
The Act includes relief for small and mid‑size enterprises, but the "we didn’t know" defense does not apply once the rule has been in force for weeks. Proactive compliance is the safest path.
Conclusion
Article 50 forces a simple but non‑negotiable change: every bot that could be mistaken for a human must announce its AI nature from the first interaction. The technical effort is minimal – a single sentence and a persistent badge – but the compliance risk is high. By updating your WordPress, Discord, and Telegram bots today, you avoid costly fines and reinforce user trust.
See how Aidevelopia answers your customers while staying compliant. Explore the demo.