01Art. 50(1)
Chatbots and AI assistants
We recognise 60+ chat tools (Tidio, Intercom, Chatbase, Botpress, Smartsupp, Infobip, Userlike, moinAI…) and look for text telling visitors they are talking to an AI.
Free · Art. 50 applies since 2 Aug 2026
Does your website comply with Article 50 of the AI Act?
Since 2 August 2026, visitors must know when they are talking to an AI, and deepfakes must be labelled. Enter your website address: we find chatbots, AI assistants and AI images, and check whether the required disclosure is there.
Scan result
- Page
- AI and chat tools found
- Images checked
Want a detailed report and a fix plan?
We also review what the scanner cannot see (the chatbot's first message, subpages, the blog) and send you a list of concrete changes.
Automated check of one page, not legal advice.
What we check
Four things Article 50 asks of a website
02Art. 50(2), 50(4)
AI images and C2PA
We look in image metadata for C2PA credentials, the IPTC "trainedAlgorithmicMedia" value and traces of tools such as Firefly, DALL·E and Midjourney.
03Art. 50(4)
Visible AI labels
We look for labels such as "AI-generated" or "created with AI" next to content made by AI.
04good practice
AI policy
We check for a link to a statement on how you use AI. It is not a legal requirement, but it is the first proof to a regulator and to customers that you have thought about it.
What the law says
Article 50 in three sentences
Chatbots: people must be told they are interacting with an AI system, at the latest at the first interaction, unless it is obvious from the context. A note hidden in the terms of use is not enough.
Deepfakes: AI-made images, audio or video that realistically depict real people, places or events must be visibly labelled. The same applies to AI text that informs the public on matters of public interest, unless it went through genuine editorial review.
Machine-readable marking: providers of generative AI systems must mark their output (e.g. C2PA, watermarks). For systems on the market before 2 Aug 2026, the deadline is 2 Dec 2026.
2 Aug 2026
Article 50 applies
€15M
or 3% of annual turnover: the maximum fine for breaching Article 50
2 Dec 2026
marking deadline for existing generative systems
Limits
What the scanner cannot see
- We do not run JavaScript. A chat loaded through Google Tag Manager or after cookie consent may not be found.
- We cannot see the chatbot's first message, which is usually set in the vendor's dashboard. Check it by hand.
- We check one page and up to 6 images. Many CMSs strip metadata when optimising images, so an unmarked image can still be AI.
- The result is a technical check, not legal advice. Whether something is a deepfake or "obviously AI" depends on context.
Questions
Frequently asked questions
Yes. The AI Act has no size exemption for transparency obligations. For SMEs the fine is the lower of the two amounts (fixed sum or share of turnover).
Article 50(1) formally binds the provider of the AI system, but a chatbot on your site, under your name and with your settings, is in practice your responsibility. The simplest fix is a first message that clearly says it is an AI assistant.
Short and at the start of the conversation, e.g. "Hi! I'm Company X's AI assistant. Type 'agent' to talk to a person." It should be in the visitor's language and visible before the first answer.
A visible label is required for deepfakes: content that realistically depicts real people, places or events. For obviously illustrative or artistic images the duty is lighter, but a label and preserved C2PA metadata are the easiest way to avoid doubt.
Croatia's implementing law is still being drafted and its authorities are not finally designated, but Article 50 applies directly as an EU regulation. In Germany the central authority is the Bundesnetzagentur; in Austria, the KI-Servicestelle at RTR.
We store only the domain, the time and the overall result, for statistics. We do not store your raw IP address or the page content.
Register interest
Monitoring and fix pack
Coming soon: monthly automatic monitoring of your site and a ready-made fix pack (chatbot AI disclosure, AI image labels, AI policy template). Sign up and you also get a free manual review of one page.