How can it be possible to find out if an image or video is AI-generated?

AI-generated media often contains subtle artifacts and inconsistencies that can be detected by specialized tools. These tools identify patterns reminiscent of synthetic images, helping distinguish them from real photos.

How accurate is the AI image detector?

In an independent benchmark by researchers at the University of Rochester and the University of Kansas using 80,000 images across multiple generators, our detector achieved the highest accuracy among the tools tested. Accuracy depends on the generator, image quality, and post-processing; no detector is 100% reliable, and results should be combined with human judgment.

Does it work without metadata or watermarks?

Yes. Detection works on the pixel content of the media itself. It does not rely on EXIF metadata, C2PA provenance tags, or visible watermarks, all of which are commonly stripped when images are uploaded to social networks, messaging apps, or marketplaces.

Which AI image generators can it detect?

The detector covers all major current generators including Nano Banana, GPT 2, Grok Imagine, Midjourney, DALL-E 3, Imagen 3 and 4, Stable Diffusion (SD 1.5, SDXL, SD3), Flux, Ideogram, Adobe Firefly, Bing Image Creator, and GAN-based face generators. The detector is updated continuously as new generators are released.

Can it detect deepfakes and face swaps?

Yes. Alongside AI image & video detection, we offer dedicated deepfake detection that identifies face swaps and AI-driven face manipulations. Both run on the same upload, so you receive a combined assessment.

Is this the same as a deepfake detector?

They overlap but are not identical. AI detection identifies any image/video generated or substantially edited by AI. Deepfake detection specifically identifies face swaps and facial manipulation, which is a narrower task. We provide both, and most users benefit from running them together.

How do you handle new AI image & video generators?

We monitor new generators continuously and add them to the detector typically within a few weeks of release. There is usually a short window after a model launches when detection accuracy on its outputs is lower than for established generators.

Can AI media detectors replace human judgment?

While the AI media detector provides real-time insights, it should never be used as a substitute for human judgment. Different detectors can have a varying levels of accuracy. It is essential to use your own discretion when reviewing their findings.

What are the limitations of AI media detectors?

AI media detectors can produce inaccurate results: false positives or false negatives. This detector may also struggle with content produced by AI media generators it has never encountered before, such as newly released ones. In some cases, images may also have been heavily edited in order to try to evade detection.

What does the provided score mean?

The score indicates the API's confidence in its prediction. A higher score suggests a greater likelihood that the content is AI-generated or AI-edited, while a lower score indicates the opposite. Even if the API returns a very high or very low score, such as 99% or 1%, it is important to consider the context and use human judgment in the final assessment.

How do you take into account new AI media generators?

We frequently test the API against new media generators as they enter the market, and take them into account to update the detectors to recognize them. Please note there usually is a brief window between the release of a new generator and its full integration into our system.