Article50.io pricing: free Article 50 scan and €499 one-off report
Last updated:
Article50.io is an automated scanner for EU AI Act Article 50 transparency gaps on public web pages, from article50.io. The free quick scan loads the page you enter in a real browser, with no signup, and shows the most severe possible Article 50 disclosure gap it finds (one finding per site every 30 days). The €499 one-off report lists every finding, with fixes.
Every full report carries a unique report ID and a SHA-256 fingerprint on a public check page, so anyone can check the file is unaltered. It is an automated technical check, not legal advice.
| Compared | Free scan | One-off report |
|---|---|---|
| Price | €0 | €499, one-off |
| What runs | Automated scan of one public page in a real browser, capped at 45 seconds | The same scan, written up as a report |
| What you get | The most severe finding, on screen | Every finding, in a dated PDF with a unique report ID, remediation steps and paste-ready disclosure wording |
| Tamper-evident file | No | SHA-256 fingerprint shown on our public check page |
| Recheck | Run it again any time: within 30 days, repeat free scans of the same site show how many issues remain and whether the first finding is still detected | One free recheck within 30 days |
| Subscription | None | None |
Free scan
- Price
- €0
- What runs
- Automated scan of one public page in a real browser, capped at 45 seconds
- What you get
- The most severe finding, on screen
- Tamper-evident file
- No
- Recheck
- Run it again any time: within 30 days, repeat free scans of the same site show how many issues remain and whether the first finding is still detected
- Subscription
- None
One-off report
- Price
- €499, one-off
- What runs
- The same scan, written up as a report
- What you get
- Every finding, in a dated PDF with a unique report ID, remediation steps and paste-ready disclosure wording
- Tamper-evident file
- SHA-256 fingerprint shown on our public check page
- Recheck
- One free recheck within 30 days
- Subscription
- None
What the free scan checks
The free scan opens the page you enter in a real browser, waits for it to load, and clicks nothing. It doesn't crawl the rest of the site. Each scan is capped at 45 seconds. It looks for:
- →Undisclosed AI chat, voice and avatar widgets. It reads text inside chat-like frames among the first 25 frames on the page, including frames from the vendor's own site, and in open shadow roots, as the widget first loads. It recognises AI disclosures in English, German, French, Spanish, Italian, Dutch and Portuguese.
- →How clearly disclosures are shown. Where the scan finds a disclosure, it checks that the AI disclosure isn't hidden or tiny (under 10px), flags very low contrast (below 3:1), and checks it isn't hidden from screen readers via ARIA. Anything it can't measure, such as text inside a frame from another site, is reported as not determined, never as a problem.
- →AI-marked images without a visible label. After the browser closes, it reads the start of up to 8 images, the social-share image first, then by size. It looks for an IPTC digital source type or a C2PA manifest saying the image is AI-generated, and flags the image only if no label, caption, alt text or other page text says so.
You then see the most severe finding. A repeat free scan of the same site within 30 days shows how many issues remain, not the next finding. A missing image marker is never a finding.
Full method: how the scan works.
What the €499 report includes
- →Every finding the scan detected, not just the most severe.
- →A dated PDF with a unique report ID. The ID appears on the PDF, in the report email and on the report's public check page.
- →A SHA-256 fingerprint on our public check page. Anyone can pick the PDF on that page and see whether it matches the report exactly as issued. The browser computes the fingerprint locally and the file is never uploaded. The fingerprint shows only that the file hasn't changed since it was issued. It says nothing about whether the site is compliant.
- →Remediation steps for each finding.
- →Paste-ready disclosure wording for your chat, voice or avatar widget.
- →One free recheck within 30 days, to confirm your fixes show up in a new scan.
- →An embeddable "Checked for Article 50 Transparency" badge.
It is a one-off payment; there's no subscription. See the refund policy.
Who it's for
- →Teams running a website chatbot, AI voice agent or AI avatar for EU visitors, who need to check whether visitors are told they are dealing with an AI.
- →Publishers and marketers using AI-generated images, who want to see whether marked images carry a visible label.
- →Agencies and consultants who want a dated, tamper-evident file to share with a client.
What the scan can't see
- →Other pages. The scan reads only the page you enter, so scan each key page separately.
- →Anything behind a click. Nothing on the page is clicked or opened, so a widget that only appears after a click or a login isn't seen. The scan sees the page as a first-time visitor does before accepting cookies. A chatbot that loads only after cookie consent won't be detected, so check it yourself.
- →Text it can't read. Closed shadow roots and canvas-drawn text aren't read, and only the first 25 frames on the page are considered.
- →How a disclosure in a cross-site frame is presented. The scan reads that text but doesn't measure its size, contrast or position.
- →Audio. For voice widgets it reads what the widget shows on the page; it can't hear what is said on a call.
- →Image labels in other languages. The scan matches visible-label wording in English only and reads only up to 8 images.
- →Watermarks and your own generator's marking. The scan doesn't detect watermarks or check Article 50(2) marking compliance.
- →Legal scope. It doesn't decide whether you are a provider or a deployer, or whether an exception applies.
Article 50 dates
- →2 August 2026: Article 50 transparency rules apply, including chatbot disclosure under Article 50(1) and deep fake disclosure under Article 50(4).
- →2 December 2026: providers of generative AI systems placed on the market before 2 August 2026 must meet Article 50(2) machine-readable marking (Article 111(4)).
Where else Article50.io is listed
Article50.io is listed in the independent Article50.tools directory at article50.tools/tools/article50-io, under Governance, tagged Freemium. The listing describes it as an automated scanner that loads a public website in a browser and flags possible Article 50 transparency gaps: chatbots, voice agents and AI avatars without an AI disclosure, and AI-marked images without a visible label. It maps the tool to Article 50(1) and Article 50(4), and says: "Free scan; a one-off paid report adds suggested fixes and a recheck."
Contact
Questions about the scan or a report: [email protected].
Not legal advice. Article50.io is an automated technical check of what a browser sees on your public pages. It doesn't give legal advice and doesn't certify compliance. For legal questions, ask a qualified lawyer.
Frequently asked questions
How much does Article50.io cost?
Article50.io has two prices. The quick scan is free, needs no signup and shows the most severe possible Article 50 disclosure gap on the page you enter. The full report costs €499, paid once, with no subscription, and lists every finding with a ready-to-paste fix and one free recheck within 30 days.
What does the free scan check?
It loads the page you enter in a real browser, clicks nothing and looks for AI chat, voice and avatar widgets without an AI disclosure, how clearly disclosures are shown, and AI-marked images without a visible label. It shows the most severe finding. It scans that one page, not the whole site.
How long does a scan take?
Each scan is capped at 45 seconds, so results usually appear in under a minute. The scan loads the page you enter in a real browser, waits for it to load and clicks nothing. Image checks run after the browser closes and share the same time limit, so a slow page never makes the scan run longer.
What is in the €499 report?
Every finding the scan detected, in a dated PDF with a unique report ID, with the Article each relates to, suggested disclosure wording you can paste in, remediation steps, a map of what the scan could and couldn't check, and one free recheck within 30 days. A SHA-256 fingerprint of the PDF is shown on our public check page.
What does the SHA-256 fingerprint prove?
Only that the PDF hasn't changed since it was issued. Anyone you share the report with can pick the file on our public check page; their browser computes the fingerprint and the file is never uploaded. It says nothing about whether your site complies with Article 50, and reports issued before 30 October 2026 have none.
Does a clean scan mean my site complies with Article 50?
No. The scan is an automated technical check of what a browser sees on one public page. It can't see content behind clicks, logins or consent choices, it doesn't decide whether you are a provider or a deployer, and it isn't legal advice. Use the report as a starting point and ask a qualified lawyer.
When do the Article 50 rules apply?
Chatbot and deep fake disclosure under Article 50(1) and 50(4) apply from 2 August 2026. Providers of generative AI systems placed on the market before 2 August 2026 have until 2 December 2026 for Article 50(2) marking, under Article 111(4).