An AI content detection platform that estimates the probability an image or text was generated by AI, identifies likely generator models, and highlights localized regions that appear synthetic. It combines computer vision algorithms for images and NLP for text, with a basic free tier covering image and text classification. Paid plans add localized detection, model identification, and API access (Basic $10/mo with 10,000 requests/day; Pro $30/mo with 40,000 requests/day; Enterprise custom). The browser extension is still listed as “in development.” Critically, independent testing has raised serious accuracy questions: Debunk.org (Oct 2026) found Illuminarty caught only 4 of 10 AI images, failed to analyze six real photos over 3.5MB with no error message, scored a 2006 real photo as 51.9% AI, and showed wildly inconsistent scores (87.7% for Midjourney vs 3% for Flux). A CBC test (Mar 2025) also found it got all three test cases wrong. The site publishes no accuracy figure and includes a disclaimer of “no representations or warranties” about completeness or accuracy.
Pros & Cons:
Pros:
Provides both image and text detection in one tool, covering two common AI-content verification needs; paid tiers add localized region highlighting and model attribution, which go beyond a simple probability score; API access is available on paid plans with documented request limits; basic detection is genuinely free and the vendor states it will “always be free”; an academic paper (arXiv 2024) included it among commercial detectors studied, giving it some research visibility.
Cons: Independent accuracy testing is damning—Debunk.org found it caught only 4/10 AI images and misclassified real photos; CBC found it got all three test cases wrong; the vendor publishes no accuracy figure and disclaims warranties; it failed to process large images (>3MB) with no error message in testing; scores are wildly inconsistent across generators (87.7% vs 3%); the browser extension is still “in development” with no release date; the vendor itself acknowledged in 2023 that the detector “tends to pick up some of the drawing tools… as artifacts coming from an AI generator,” meaning digital artists may be falsely flagged.
Who is it for?:Users who want a free, quick probability estimate on an image or text and understand it is a signal, not a verdict; researchers or journalists who need an API for bulk detection and will cross-check results with other tools; users verifying AI-generated content where a false positive is tolerable and human review follows.