Skip to content

What do you want to check?

Choose the way that fits what you received.

Detection runs in your browser. Nothing is sent until you start the check.

Upload a screenshot instead

  • Email

    You need
    The suspicious email, to forward or as an .eml file.
    You get
    A report with a verdict, what stands out about sender, links and attachments, and the next steps.

    Time: In our measurement on 2 October 2026 the analysis took 25 to 40 seconds. Delivery of the email to us can take longer, sometimes over a minute.

  • SMS & messengers

    You need
    The text of the message, or a screenshot of it.
    You get
    The same report as for emails. For screenshots your browser recognises the text, and you can correct it and black out passages.

    Time: The report appears as soon as the analysis is done. For your first screenshot the browser downloads about 8 MB of language data, once.

  • Link

    You need
    The link or domain from an email, text or ad.
    You get
    Whether the address is in phishing lists, regulator warnings or our own records, and what stands out in how it is written.

    Time: Right on the page, without us opening the link. No entries is not an all-clear.

  • Phone number

    You need
    The phone number that called or wrote to you.
    You get
    Number type, notes on ping-call prefixes and premium numbers, and a comparison with Federal Network Agency measure lists and our own reports.

    Time: Right on the page. The displayed number can be spoofed.

  • IBAN

    You need
    The IBAN you are supposed to pay or have paid.
    You get
    An answer on whether approved fraud reports exist for this IBAN.

    Time: Right on the page. No report is not an all-clear.

Online shops do not belong here. Check an online shop at fakeshop.ai sister service

How we check

Every check states what it is based on. Rules and lists decide; a language model only helps select passages.

AI text. Text generated by AI, without human review in the individual case, Explanation: Where we use AI
Email
We read the original’s headers and check whether the sender matches the domain. Links and attachments are matched against warning lists. A language model marks passages that stand out; the explanation is a fixed text of ours.
Texts and messengers
The same report as for emails, without headers. A text’s sender can be forged, so the content counts most: links, deadlines, payment demands.
Link
We read only the domain of the link, without opening it, and match it against phishing lists, regulator warnings and our own records. In the spelling we look for short links, special characters, brand names in unrelated domains and IP addresses.
Phone number
We determine the country and type of number and recognise prefixes common in ping calls, then match it against the Federal Network Agency’s list of measures and reports sent to us. No language model is involved here.
IBAN
Your browser checks structure and check digits. The server looks up whether approved fraud reports exist for the account. We do not check who owns an account.
Sender address
We check whether the domain exists, whether it can receive email, how old it is and whether it imitates a known brand. We also match it against financial regulators’ warning lists, a phishing list and findings from our email checks. We never query the remote mail server.
Password
Your browser computes a checksum of the password and sends us five characters of it. We fetch from Have I Been Pwned the list of all entries with this beginning; your browser does the comparison. We learn neither the password nor the result.
What the check cannot do
It is an automated assessment and replaces neither your bank nor the police. New numbers, accounts and addresses are often not on any list yet. A result without findings is therefore not an all-clear.

The diagram shows how a check works in three steps, from left to right.

  1. What reaches you: Email; SMS and messenger; Phone call; IBAN in an invoice.
  2. What we compare it with: Warning lists of financial regulators, BaFin, AMF; Phishing lists, active domains; Number plans of regulators, DE and AT; Bank codes of the Bundesbank; Sender checks, SPF, DMARC, domain age; Reports from people affected; Text check, by a language model.
  3. What you get: The verdict with the marked parts; The reasons with source and date; The next steps Block, report, file a complaint.

Example: the phrase “pay today, please” in a message is marked yellow during the comparison and appears in the verdict.

Marked yellow: a conspicuous phrase in the text. The example is made up.

Automated assessment, partly generated with AI. No guarantee, not legal advice. Liability as set out in our terms of use. How we check.

Our data sources

Besides the reports sent to us, we import open data from authorities and other operators. Each source is listed with its operator, purpose, licence and date.

What happens to my data?

You only enter what you want checked. We use it for the report and not for advertising.

Emails are deleted
We delete the forwarded original email after the check.
Report for seven days
You can open the report through your link for seven days.
Screenshots stay with you
Text recognition runs in your browser. The image never leaves your device, you see the text and can edit it before sending.
The IBAN never appears in a URL
The IBAN travels in the request body to our server and is not stored there. Your browser verifies its check digits beforehand.
AI use is labelled
A language model picks out passages of your message that stand out. The explanations are fixed texts written by us. This label under Article 50 of the EU AI Act shows where it took part.AI text. Text generated by AI, without human review in the individual case, Explanation: Where we use AI