Voice assistant
Article 50 asks that visitors know they are talking to an AI. Scan your site free, and read below exactly where that notice goes in ElevenLabs ConvAI (voice).
Free · No account · Result in ~30s
ElevenLabs Conversational AI sounds convincingly human, and that is exactly why this page differs from a chat widget one. Article 50(1) carries an exception for situations where it is obvious to a reasonably observant person that they are talking to an AI. With a voice indistinguishable from a person, you cannot lean on that.
We have not yet scanned enough sites where ElevenLabs ConvAI (voice) was the first widget we detected to say anything meaningful here. As soon as the sample is large enough it will appear, counts included. We would rather publish nothing than a percentage based on two sites.
Our scanner looks for these signals on your public page. Two independent signals are required for a match, so a blog post that merely names the vendor does not count.
api.elevenlabs.iounpkg.comnoneelevenlabs-convaiThe first sentence the agent speaks, the opening message field in the agent config. A text banner is not enough here.
Open your agent in the ElevenLabs dashboard.
Go to the agent configuration and find the first message field.
Put a short spoken AI notice there, as the opening line.
Embedding the widget on a page? Put the sentence as visible text beside it too.
Call or start the agent yourself and listen for the line actually coming first.
The sentence you paste
You are chatting with an AI assistant, not a person.
The welcome message only appears after opening. The banner snippet from the Compliance Kit covers the moment before that with a single script tag, and your AI transparency statement covers the question that comes after.
Watch out with ElevenLabs ConvAI (voice): This is voice. The notice must be audible before the user says anything, not halfway through and not only on the page.
ElevenLabs ConvAI (voice) documentation: elevenlabs.io
Three mistakes we run into on sites that did set the notice. Each one costs points in the scan, and each one takes minutes to fix.
Instructing the agent to admit being an AI when asked is not a notice. The visitor has to hear it without asking. Put the sentence in the first message field, not in the prompt.
With barge-in enabled the caller can talk over the opening line. Keep the notice short and put it right at the front, so it has already landed even when interrupted.
An agent introducing itself as "Julia" in a natural voice is precisely the case where the Article 50(1) exception does not hold. Add in the same sentence that Julia is an AI assistant.
This check is one you can only run yourself, because a scan hears nothing. Start a conversation with your own agent and say nothing: the notice should sound before it is your turn. Do it again and start talking immediately, to see whether barge-in cuts the line short. Finally ask literally whether you are speaking with a person and listen for the answer to confirm. Record the call and keep it with your AI transparency statement.
After that you can scan again. Our scan looks at the same public page a visitor does, so what you see in an incognito window is roughly what we see.
It matters more. The more convincing the voice, the less obvious it is that this is AI, and the Article 50(1) exception only applies when it is obvious.
Every conversation, yes. Every sentence, no. One short opening line per conversation is the common reading of "at the latest at the first interaction".
No. Our scan reads your public page and recognises the widget embed. What the agent says, we cannot hear. For voice, your own test call is the only real check.
The Compliance Kit contains a copy-paste disclosure banner (NL/EN/DE), an AI transparency statement, an AI system inventory and a step-by-step guide per widget.
See the Compliance Kit (€149)Technical compliance tooling and templates. Not legal advice.