Password Breach Checker

Checks Have I Been Pwned without ever sending your password. Only five characters of its hash leave your browser.
Client-side · verifiable
How this tool works

Have I Been Pwned maintains a corpus of more than 900 million passwords found in real breaches. Checking against it without revealing your password is possible thanks to a technique called k-anonymity:

  • Your browser computes the SHA-1 hash of the password, 40 hex characters.
  • Only the first five characters are sent to api.pwnedpasswords.com/range/.
  • The API returns every hash suffix in the corpus that starts with those five characters, typically 800 or so, padded with fake entries so the response size reveals nothing.
  • Your browser searches that list for the remaining 35 characters. The match, or the absence of one, is found on your device.

The server learns that someone asked about a bucket of hundreds of possible hashes. It cannot know which one, or whether any matched. Hashword never sees the request at all: it goes straight from your browser to HIBP.

Security note

What leaves your browser

The password never leaves your browser. The hash is computed locally with the Web Crypto API and only a 5-character prefix is sent, directly to Have I Been Pwned, with the Add-Padding header so response sizes leak nothing. Hashword has no server component in this check.

FAQ
Is it safe to check my real password?

Yes. This is the same mechanism 1Password, Bitwarden and browsers use for their breach alerts. The full hash never leaves your device, and a 5-character prefix matches hundreds of unrelated hashes.

My password was found. What now?

Change it everywhere you used it, most important accounts first. Turn on two-factor authentication. Then read our guide on what to do after a data breach.

It was found once. Does that matter?

Yes. Any password in the corpus is in attackers' wordlists. Cracking tools try the whole corpus before anything else.

Not found means safe?

It means not known to be leaked. A short or predictable password can still be guessed. Run the strength checker too.

Why does it use SHA-1 if SHA-1 is broken?

SHA-1 weaknesses concern collisions, which are irrelevant here. The HIBP corpus is indexed by SHA-1, so the lookup must use it.