TruthLens

What TruthLens is

TruthLens is a free, open-source browser extension that flags possible AI-generated text, synthetic images, and cloned voices. Its text checks use patterns such as burstiness, common-word ratios, phrase fingerprints, and syntactic uniformity. Image checks apply pixel-level heuristics and look for known AI image sources, including Midjourney, DALL·E, Stable Diffusion, Leonardo, and Ideogram. Audio and video checks examine spectral flatness, silence regularity, and pitch jitter. Clickable warning badges explain which signals contributed to a flag. You can rescan with Ctrl+Shift+T or right-click an image to triage it. The maker says detection runs on-device, with no uploads, accounts, or tracking. It cautions that TruthLens is a triage aid, not forensic proof, and that its signals can produce false positives and false negatives. The repository lists compatibility with Chromium-based browsers including Chrome, Edge, Brave, Opera, Arc, and Vivaldi. Installation involves downloading or cloning the repository and loading it unpacked in the browser’s extensions page. A signed Chrome Web Store release is planned after the public beta. The maker names journalists, teachers, researchers, parents, voters, and recruiters as intended users. TruthLens is licensed under MIT.

Who it suits

TruthLens is intended for people such as journalists, teachers, researchers, parents, voters, and recruiters who want signals for triaging potentially synthetic content. It requires a listed Chromium-based browser and manual unpacked-extension installation.

Limits

TruthLens is a triage tool rather than forensic proof, and its detection signals can produce false positives and false negatives.

Getting started

1. Open the TruthLens project website or repository
2. Clone or download the repository
3. Open your Chromium-based browser’s extensions page
4. Load the repository as an unpacked extension
5. Browse supported pages and use inline badges, Ctrl+Shift+T, or image right-click triage

Verdict

TruthLens provides multiple kinds of detection signals and explains the signals behind its warnings. Treat results as triage rather than proof, and note the manual installation process and stated error risk.