Stop repeat fraud and abuse without adding login friction.
Know when the same physical device returns—even across browsers, cleared cookies, or private windows—so you can block multi-accounting, catch stolen credentials faster, and step up MFA only when risk is real. Stable IDs in milliseconds, not another hoop for good customers.
Signals per identification
p95 end-to-end latency
Major browsers, one device ID
Tamper & API checks
Capabilities
Everything you need
DRAWNAPART GPU Timing
Inspired by DRAWNAPART research, our GPU timing analysis uses 10K+ vertex workloads to extract hardware-specific rendering signatures that distinguish individual GPUs.
Attention-Based ML Matching
Neural attention model weighs 37+ signal dimensions to produce a confidence-scored device match, continuously improving through our real ML training pipeline.
Canvas Fingerprinting
Sub-pixel rendering differences in HTML5 Canvas create unique signatures based on graphics hardware, OS, and font rendering engines.
AudioContext Analysis
Audio processing characteristics of the device's sound hardware produce consistent, cross-browser identifiers.
Cross-Browser Consistency
The same device ID is generated whether the user visits from Chrome, Firefox, Safari, or Brave. Our core differentiator.
Incognito Persistence
Hardware-level signals don't change between normal and private browsing sessions, making our fingerprints persistent across modes.
Signal Drift Detection
Automated monitoring detects when GPU, WebGL, or audio signals drift over time due to browser updates, ensuring accuracy as the landscape evolves.
Lies & Tampering Detection
Scans 197+ browser APIs for prototype tampering, spoofed properties, and fingerprint resistance — detecting manipulation attempts in real time.
Headless & Automation Detection
Identifies headless browsers, Puppeteer, Playwright, and automation frameworks through engine-specific signal analysis.
Behavioral Biometrics
Captures mouse dynamics, keystroke timing, scroll patterns, and touch gestures to strengthen identification and detect bot-like behavior.
Sub-40ms Latency
Signal collection, hashing, and server-side matching complete in under 40 milliseconds. Zero impact on page load times.
How It Works
Three steps to get started
Install the SDK
Ship identification in one dependency. Full TypeScript types so your team wires sessions and user IDs with compile-time safety.
Call identify()
On each visit, the SDK collects GPU, canvas, audio, WebGL, tamper checks, and behavioral signals—no prompts, no extra clicks for legitimate users.
Act on the verdict
Use device ID, confidence, and risk in your rules engine: allow, challenge, or block—typically in under 40ms end-to-end.
Live Playground
Your device, identified right now
The JSON on the right is the actual fingerprint just collected from your browser — no fixtures, no mocks. Switch languages to see the call that produces it.
import { LightningResearch } from "@lightningresearch/sdk";
const { sessionToken } = await fetch("/api/collector/session", {
method: "POST",
}).then((r) => r.json());
const client = new LightningResearch({
sessionToken,
endpoint: "https://api.lrdefender.lightningresearch.ai",
});
const result = await client.identify();
// → { deviceId, uniqueness, elapsed, signals, hashes }
Use Cases
Built for real-world security
Payment & signup fraud
Link chargebacks and stolen cards to the same hardware even when fraudsters rotate emails, browsers, and incognito windows—so risk teams review fewer false leads.
Trust scoring at login
Treat unrecognized devices as higher risk and trigger MFA or device binding only when Trace says the session looks new or inconsistent—not on every visit.
Promo & trial abuse
Stop one person from farming referral credits, trials, or signup bonuses by detecting repeat devices behind fresh accounts.
Marketplace & seller integrity
Surface sock-puppet sellers and coordinated fake buyers by resolving multiple storefront logins to the same device footprint.
High-value account recovery
When a user resets a password or moves money, compare the requesting device to history to block takeover flows that look like new hardware.
Get started with LR Trace
Free tier includes 10,000 identifications per month. No credit card required.