Keystroke dynamics
Records key press and release times to build a typing rhythm profile.
In one line
Can time your key presses to learn your typing rhythm
Records key press and release times to build a typing rhythm profile.
Dwell times (key held) and flight times (between keys) for password and free text identify people with 1 to 5% error. Persists across devices and accounts, used for continuous authentication and fraud.
What it can and can't do
What the user does: You type in a box on the page
It can
- Time how long you hold each key
- Time the gap between keys
- Build a typing profile that picks you out with about 1 to 5% error
- Match the same person on other devices or accounts
It cannot
- Work if you paste text instead of typing it
- Ask permission first
- See what you type in other tabs or apps
- Be switched off by clearing cookies
Browser support
| Browser | Status | First version | Released | Note |
|---|---|---|---|---|
| Chrome | Supported | 1 | 2008-12-11 | None |
| Edge | Supported | 79 | 2020-01-15 | None |
| Firefox | Supported | 1 | 2004-11-09 | None |
| Safari | Supported | 1 | 2003-06-23 | None |
| Safari on iPhone | Supported | 1 | 2007-06-29 | None |
| Chrome on Android | Supported | 1 | Not listed | None |
| Samsung Internet | Supported | 5.0 | 2016-12-15 | None |
How a site asks for it
| API | keydown/keyup events with event.timeStamp; InputEvent |
| How access is granted | none |
| What the user sees | Not listed |
| Needs a secure context (HTTPS) | No |
| Needs a user gesture | No |
| How long it lasts | Not listed |
| How to take it back | Not listed |
| Inside an iframe | Not listed |
Typical flow
- Listen to keydown and keyup on fields
- Record timestamps per key class
- Compute dwell and digraph latencies
- Compare with enrolled profile
Data you get
What comes back: Timestamps at 0.1 to 1 ms
Shape: list
| Field | Type | Unit | Range | Precision |
|---|---|---|---|---|
dwell | float | ms | 40 to 250 | timer resolution (0.1 to 1 ms) |
flight | float | ms | -100 to 600 | same |
Technical limits
- Cannot read keys outside the page or in other iframes
- Password fields still emit events to the page
The story
What it is
A script records when each key goes down and up, to build a profile of your typing rhythm.
For you
You never see it. Everyone types with their own rhythm, like a signature. A page that can see key timing can learn yours and later tell if it is you or someone else typing.
For makers
Banks and fraud teams use typing rhythm as a quiet second check that you are really you. Some sites use it to spot bots that type too evenly. Critics point out it is also tracking of a body trait.
Where you see it
| Type | Example | Apps |
|---|---|---|
| banking | A bank compares how you type your password with your usual rhythm, and asks for more proof if it looks different. | None named |
| security | A bot detector sees login typing that is too even and fast, and blocks it as a script. | HUMAN, DataDome |
| education | An online exam service checks that the same person who started the course is the one typing the test. | Coursera |
| enterprise | A company portal uses typing rhythm as a hidden second factor for staff accounts. | None named |
| social | A social site spots a bot farm because all fake accounts type with the same machine like pattern. | None named |
| games | A browser game with chat checks for macro or script input that types perfect timing. | None named |
| ads | A tracker could link two visits by typing style, which is why privacy groups raise concerns. | None named |
Risks
- Typing style is personal
- Hard to notice or avoid
- Can be used for tracking
Good uses and abuse
Legitimate uses
- Fraud detection, account takeover defense, typing tests
Known abuse
- Cross-device identity tracking and profiling, user re-identification
Scores
| Score | Value | Meaning |
|---|---|---|
| Sensitivity | 3 of 5 | How revealing the data is |
| Utility | 4 of 5 | How useful it is to honest apps |
| Friction | 0 of 5 | How much the user has to do |
| Fingerprint bits | 4 | Rough identifying entropy, when known |
History
- 2013-01, paper: Monrose and Rubin keystroke authentication line of work; Killourhy and Maxion 2009 benchmark
Related capabilities
Related
- Mouse movement dynamics. Can track how your cursor moves on the page, to tell people from bots
- Scroll and touch dynamics. Can track how you scroll and swipe, plus touch size and pressure
Sources
- Killourhy and Maxion, Comparing Anomaly Detectors for Keystroke Dynamics (DSN 2009)
- BioCatch, typing biometrics
How well checked: Based on official documentation.