Welcome!
We're building a better email security system, and we need your help.
We've all known family and friends who've fallen victim to phishing attacks, where people are tricked into providing confidential information, like login or card details, which leads to them losing money or having their identity stolen.
We have developed a system which can accurately spot malicious phishing emails, but we need much more training data, in the form of forwarded emails.
Simply forward us any emails, whether suspicious or not and our system will do the hard work.
Please remember, during this training phase it is necessary to store analysed emails, as we may occasionally need to review them to check the accuracy of system verdicts, so please don't forward us anything personal or confidential that you don't want us to see.
Cheers,
Merul & Jack
Forward any email – genuine, junk, or suspicious – to:
help@verifitrust.orgMake forwarding even easier – save us to your contacts:
Add Verifitrust to ContactsDisagree With a Verdict?
If you receive a verdict that doesn't look right, simply reply to the email (you don't need to say anything in the reply) and we'll review it manually. Your feedback is invaluable – this is exactly how we improve the model.
English Only, Please
For now, our semantic analyser is only trained in English, so please don't forward emails in other languages – the text analysis won't produce reliable results.
How It Works
We thought you might want to know more about how the system works, and there's also an FAQ that we'll keep updated.
Traditional email hosting providers do have basic spam filters which are reasonably effective but, due to the large volumes of email they have to process, they can't check individual emails exhaustively. As a result some junk emails or phishing emails end up in inboxes, and some genuine emails end up in junk. And in any case, we all check our junk folders from time to time and have to spend time determining whether they are genuine or malicious.
No installation, no setup – just forward an email and we do the rest
Forward the Email
Forward any email – suspicious or not – to help@verifitrust.org. We need genuine, junk and phishing examples to improve the model.
We Analyse & Scan
Our system performs multiple checks on the email before reaching a verdict:
- ● Virus scanning – attachments are checked for malware, just in case your ISP missed something
- ● Sender investigation – we look up who owns the sender's domain and the reply-to domain, and when they were created
- ● URL analysis – every link embedded in the forwarded email is extracted, compared to the sender's domain, then followed to its final destination to see where it really leads
- ● Invoice detection – any attachments are scanned for bogus invoices and any details are extracted for review
- ● Over 50 other features analysed for each email
AI Analysis
Multiple machine-learning models work together to classify the email:
- ● Semantic analysis – a neural network analyses the text, including content rendered in a simulated browser and text read from images via OCR
- ● Invoice detection – a large language model extracts invoice details from any attachments and analyses them for telltale signs of fraud
- ● Final verdict – a third model crunches all the gathered inputs to produce an overall verdict, together with a simple indication of our confidence level, in seconds
Receive Your Verdict
We email you back with a clear verdict – no attachments, no complex reports
Query Your Verdict
If you think the VerifiTrust system has got it wrong, simply reply to the verdict email, and we will re-analyse the original email (including in some instances inspection by a person). You will then receive a follow-up verdict, either confirming the original analysis or showing an updated verdict.
Our Latest Statistics
Based on all emails analysed so far
-
Genuine
-
Junk
-
Phishing