Technical, in plain words

How Korun works

Korun uses the URL filter Apple introduced in iOS 26 to stop fake sites without a VPN and without seeing your browsing. Here is what happens, step by step.

1. Your phone checks first

Korun downloads a summary of known dangerous addresses (a Bloom filter). The file is the same for everyone, signed, and fetched without cookies or identifiers. When Safari or any other app opens an address, iOS compares it and its parent parts (for example login.fake-bank.xyz and fake-bank.xyz) with this summary on the phone. Most addresses pass here with no network traffic at all.

2. When in doubt, an encrypted question

If the summary says “maybe”, iOS asks Korun’s server with Private Information Retrieval (PIR), a form of homomorphic encryption: the server computes the answer but can’t read the question.

3. A decision — and your internet stays on

If the address is dangerous, iOS stops the request. If Korun’s server can’t be reached, iOS opens the site; Korun tells you the shield needs attention and lets you reinstall it with one tap. Protection never cuts your connection.

4. Safari tells you why

For a blocked page iOS normally just says “Safari can’t open the page”. With Korun’s Safari extension turned on, the blocked address is analysed on your phone and you see a plain explanation such as “Stop. This site is a fake PTT. The real address is ptt.gov.tr.” The extension never reads page content.

5. Where the lists come from

Dangerous-address lists are compiled from public sources and our own rules on our server, then signed and published. Nothing goes from your device to those sources. While compiling:

6. Checks happen on your phone

Messages, links, e-mails, PDFs and screenshots you share are analysed only on your phone. Korun Advisor uses Apple’s on-device language model on devices with Apple Intelligence; content never leaves the phone. The verdict always comes from explainable rules — the model only writes the explanation.

What it doesn’t do