Frequently Asked Questions

An AI image detector analyzes a photo's visual patterns, textures, and pixel-level signals to estimate whether it was created or modified by artificial intelligence tools like Midjourney, DALL·E, or Stable Diffusion.

Yes — the Free plan checks 15 images a month at no cost, no card required. Paid plans raise every monthly allowance and add scan history, batch uploads, scan-by-URL, and raw JSON export.

No. Anonymous visitors get 3 free scans. A free account raises that to 10 scans every 24 hours (15 image checks a month) with shareable reports; saved scan history starts on Starter.

No detector is 100% accurate. Accuracy varies by the generator used, image compression, and whether the image has been edited after generation. We recommend using detection results as one signal alongside your own judgment.

Yes. Results depend on the generator used, image compression, and any edits made after generation. Treat results as one signal, not a final verdict, especially for high-stakes decisions.

Heavy compression, filters, or certain camera and editing software can occasionally produce patterns that resemble AI-generation artifacts. This is a known limitation of all detection tools, not just ours — it's part of why we show a confidence score instead of a flat yes/no.

Both. When a file carries embedded metadata or content credentials, we read and display it. Alongside that, we run our own pixel-level analysis that doesn't depend on metadata being present at all, since metadata is easy to strip or fake.

Images: JPG, PNG, GIF, and WebP. We also support common video (MP4, WebM) and audio (MP3, WAV, M4A) formats.

Heavy edits, cropping, and recompression can reduce detection accuracy, since they alter the signals our model looks for. Treat results on heavily edited images as a weaker signal.

When an image is flagged as AI-generated, we surface our best-guess attribution — which generator was likely used, ranked by confidence. This attribution is a secondary signal and is most reliable on unedited images.

Yes — our detector is trained to recognize patterns from generators including Midjourney, DALL·E 3, Stable Diffusion, Google Nano Banana and Nano Banana Pro (Gemini), Adobe Firefly, Ideogram, FLUX, Leonardo AI, Canva AI, ChatGPT image generation, and Grok Imagine.

Yes. Beyond photos, we also check video clips for deepfake manipulation and audio for AI-generated or cloned voices. Switch tabs on the homepage to analyze video or voice the same way you check images.

Images are used only to run the verification check. We don't sell your images. See our Privacy Policy for full details on retention and third-party processing.

Yes: up to 10MB for images, 200MB for video (60 seconds max), and 50MB for audio.

The Free plan lets you try image, video, and voice detection at no cost, with 15 image checks a month and 7-day saved history. Starter ($4.99/mo) raises that to 150 checks and extends history to 30 days. Pro ($14.99/mo) adds batch uploads, scan-by-URL, and raw JSON export. Gold ($39.99/mo) is the highest-volume tier.

Yes — nothing in our Terms of Service restricts commercial use of results.

Not yet — batch checking is on our roadmap. Today, each file is checked individually.