Pixlyze is aimed at the everyday problem of not being able to trust an image at a glance. Generative image models now produce outputs that look photographic, so editors, creators, marketers and researchers often need a second opinion before publishing or reposting something. Pixlyze handles that check in the browser: you drop in a file, the service runs it through detection models, and it returns a judgement about whether the content is likely synthetic or real.
The tool is deliberately simple and does not require an account. The main selling points the project advertises are accuracy, speed, privacy (uploaded files are said not to be stored or shared) and an interface that non-technical users can operate. Alongside detection, Pixlyze offers a counter-feature: it can process an image to make it "undetectable" and let you download the modified version, which is useful if you want AI-assisted imagery to survive automated scrutiny.
Pixlyze is a free project published by developer Vishwa Gaurav under the itsvg.in domain, not a commercial SaaS with tiered plans. It is best treated as a fast first-pass filter rather than forensic proof: no detector is infallible, and results should be combined with metadata inspection, reverse image search and human judgement before you draw firm conclusions about a photograph.
