Illuminarty AI image detector

The vendor publishes its limits as a list of known problems in the release post of 18 March 2023: "Cannot process large images (> 3MB)", with a promise to "lift the size limit of the images as we improve our infrastructure"; "Fails on specific art styles, e.g. color schemes, line styles"; "Fails on high resolution images", which the vendor attributes to "artworks that use automated brushes and automated light expressions"; and "Fails on photographs, anime and game screenshots", because such media "tend to undergo a lot of post-processing, most of which use AI algorithms for rendering and upsampling" (release post, read 9 November 2026). The web app runs at app.illuminarty.ai; the API allows 10,000 requests a day on Basic and 40,000 on Pro, with unlimited usage on Enterprise. No format list is published, and api.illuminarty.ai answered with an empty page on 9 November 2026.

Pricing is a monthly subscription with a free tier the vendor says will stay free: "AI detection will always be free, but we offer additional features as a monthly subscription to sustain the service" (illuminarty.ai/en, read 9 November 2026). The release post explains the model: the subscriptions "are made to support the growing server fees and training costs for detection". Figures below are as displayed, in dollars, on 9 November 2026; the page states no currency and the company is Swiss.

Retention is stated in one sentence. The Privacy Policy, last updated November 16, 2022, reads: "Any image uploaded to our server will be processed in order to provide our service to you" and "Any image uploaded to our server will be kept only for the duration necessary to provide our service to you" (illuminarty.ai/en/privacy, read 9 November 2026). The policy states no period in days and no deletion-after-processing sentence, so the retention rank field is empty and the clause is quoted in the data retention ranking's second table. Nothing is said about text submissions; feedback form data goes to Typeform S.L.

Illuminarty publishes no accuracy percentage on its home or image pages. Its figures are in the release post of 18 March 2023, as "validation accuracy" on "our collected dataset": binary AI detection rose from 91.8% with the previous model to 97.5% with the new one; multi-class model detection from 90.1% to 96.4%; Dalle2 from 99.2% to 99.8%; Midjourney from 68.9% to 88.5%; Stable Diffusion and its variants from 79.1% to 93.5% (release post, read 9 November 2026). The post qualifies them at once: "these numbers need not match the actual experience of the users. This is because our current collected dataset does not cover the diverse styles and tools of the human artists, the detector currently tends to pick up some of the drawing tools, e.g. automated brushes, as artifacts coming from an AI generator."

No false positive rate is published. The vendor's request to artists, "Please let us know if your works are deemed as AI generated, it helps us a lot to understand the errors", is the closest the pages come to one, and it anticipates the direction of the study result below. The terms add that "no parts of the output nor the output as whole, of the service, constitutes as a statement from the service owner or operators about any other party", and the footer disclaims "any kind" of warranty on the site's information (illuminarty.ai/en/terms, read 9 November 2026).