Skip to Content
Platform.Ontology

JA3 TLS client fingerprint

The TLS handshake order of ciphers and extensions identifies the client software.

Used to track youNetwork and connectionTLSWeb browsersWhen the page loads

In one line

Can guess which browser or app you use from how your connection starts

The TLS handshake order of ciphers and extensions identifies the client software.

JA3 (Salesforce, 2017) is an MD5 of version, ciphers, extensions, curves and point formats in the ClientHello. Chrome 110 randomized extension order (Jan 2023), so one browser now yields many JA3s; JA4 sorts them out.

What it can and can't do

What the user does: Nothing. It happens the moment your browser connects to the site

It can

  • Make a short code from the list of options your browser offers when it connects
  • Spot bots, scripts and malware that connect differently from real browsers
  • Recognize the same software again across different internet addresses

It cannot

  • See your name, account or what you browse
  • Give one steady code for Chrome, since Chrome shuffles the order and gives many codes
  • Be turned off in your browser settings
  • Tell two people with identical software apart

How it differs between platforms: Chrome and Edge shuffle the options on purpose, so one browser gives many codes. Firefox and Safari keep a steadier order.

Browser support

BrowserStatusFirst versionReleasedNote
ChromeSupported12008-12-11None
EdgeSupported792020-01-15None
FirefoxSupported12004-11-09None
SafariSupported12003-06-23None
Safari on iPhoneSupported12007-06-29None
Chrome on AndroidSupported1Not listedNone
Samsung InternetSupported5.02016-12-15None

How a site asks for it

APITLS ClientHello parsing at CDN or server
How access is grantednone
What the user seesNot listed
Needs a secure context (HTTPS)No
Needs a user gestureNo
How long it lastsNot listed
How to take it backNot listed
Inside an iframeNot listed

Typical flow

  1. Capture ClientHello
  2. Concatenate fields
  3. MD5
  4. Match against known hashes

Data you get

What comes back: 32 hex characters
Shape: string

FieldTypeUnitRangePrecision
ja3_hashstringNoneMD5exact, but unstable on Chrome 110 and later

Technical limits

  • Chrome randomization makes the hash non-stable
  • Easy to spoof with curl-impersonate or uTLS

The story

What it is

The way your browser starts a secure connection leaves a pattern that names the software you use.

For you

You never see it. Before any page loads, your browser says hello in a particular order. Servers can turn that order into a short code that points to your browser family.

For makers

Makers use it to tell real Chrome or Safari apart from scripts and bots pretending to be them. It happens before any page code runs, so it is hard to fake. Browsers now shuffle the order, which weakens it.

Where you see it

TypeExampleApps
securityA ticket site gets a flood of requests that say they are Chrome. The secure handshake code matches a Python script, so the site blocks them before the sale opens.None named
adsAn ad network checks that clicks come from real browsers. Clicks with a known bot handshake code are dropped and not billed to the advertiser.None named
gamesA browser game's login server blocks a cheating tool that automates logins, because its handshake matches a known tool and not a normal browser.None named
shoppingA sneaker store uses the handshake code to spot buying bots that fake a browser name, so real shoppers get a fairer chance.None named
analyticsA bot defence service lists the most common handshake codes in its traffic and sets alarms when a rare one suddenly appears.Cloudflare
bankingA bank's login page treats a handshake that does not match the claimed browser as a warning sign and adds a challenge.None named
developerA developer scraping a site with a plain script gets blocked, and learns the site compares the handshake code with the browser name sent.None named

Risks

  • Hard to see or avoid
  • Wrong blocks of rare browsers

Good uses and abuse

Legitimate uses

  • Bot detection, malware C2 detection

Known abuse

  • Persistent client identification across IPs

Scores

ScoreValueMeaning
Sensitivity2 of 5How revealing the data is
Utility4 of 5How useful it is to honest apps
Friction0 of 5How much the user has to do
Fingerprint bits3Rough identifying entropy, when known

History

  • 2023-01, Chrome 110: TLS extension order randomized, breaking static JA3 hashes
  • 2023-01, Firefox: Firefox keeps fixed order

Sources

How well checked: Based on official documentation.

On this page