Introduction

About Sens

Sens is a free tool for checking whether content was made by AI: images, and passages of text.


What it is for

Synthetic images are good enough that looking closely no longer works. Sens gives journalists, fact-checkers, educators, photographers and anyone with a doubt a way to ask "is this AI?" and get an answer that shows its reasoning.

How it decides

For images, a verdict is built from independent signals, ranked by how much each one can be trusted:

  1. Content Credentials (C2PA). A cryptographic signature from the generator, editor or camera that produced the file. When present and valid, it names the producer. This is the strongest evidence available, for or against AI.
  2. Generator metadata. Many AI tools write their name into ordinary metadata fields: PNG text chunks, XMP packets, the EXIF Software tag. It is trivial to strip, but rarely added to a real photo by accident.
  3. Visual detection. A neural detector (Community Forensics, CVPR 2025) trained on output from thousands of generators scores the pixels. It runs in your browser. It is probabilistic, and the verdict says so whenever it is the only signal.

The verdict shows every signal that contributed, strongest first, plus the limitations that apply. Reading the verdict explains each headline.

For text there is no provenance to lean on, so the text detector is a statistical classifier tuned to avoid false accusations, with an optional second opinion from a commercial detector.

What it never does

  • Upload your image by default. Every image check runs in your browser. The file leaves your device only if you sign in and choose to save a report. Text is different: it is sent to the server to be checked, then discarded.
  • Use the image for AI summaries. The optional summary is generated from the extracted metadata text, not the image bytes.
  • Dress up a guess as a fact. When only the visual detector has an opinion, the headline is "Likely AI-generated" or "Inconclusive", never "AI-generated".

Who builds it

Sens is made by Sens Center, Inc., a small, independent, self-funded team. The code is on GitHub, the Chrome extension is open source, and the detector is benchmarked against a labeled public corpus. Questions and wrong verdicts are welcome on Discord or through the contact form.

Previous
Welcome