AI Image and Video Detectors by Generator
Pick the generator you suspect, from ChatGPT and Midjourney to Sora and Veo, and see which provenance signals it leaves, how to check them and where checks end.
Guidance reviewed 2026-10-12. Independent service; no generator affiliation. How we review guidance
AI generators differ in what they leave behind. Some, such as Adobe Firefly and ChatGPT, attach C2PA Content Credentials to the files they export. Google's tools embed an invisible SynthID watermark. Stable Diffusion interfaces often write generation parameters into PNG files. Other generators document no provenance signal at all. Each page below covers one generator: the signals its maker documents, how to check them, and what a missing signal does and does not mean.
Start with the generator someone says made the image, or the one you suspect. If you do not know, the free checker reads Content Credentials and AI metadata in your browser without uploading the file, and the step-by-step guide covers checks that work for any image.
A generator page is not a verdict. Screenshots and re-uploads can strip provenance, and visual clues are unreliable on their own, so treat every result as one piece of evidence and ask for the original file when the answer matters.
Check by AI image generator
- ChatGPT & DALL-E Image Detector: Provenance
Was an image made with ChatGPT or DALL-E? Check OpenAI's Content Credentials and SynthID signals, use OpenAI's verifier, and learn what each signal cannot prove.
- Nano Banana Pro Detector: SynthID and C2PA
Check whether an image was made or edited with Google's Nano Banana or Nano Banana Pro: SynthID in the Gemini app, C2PA Content Credentials, and where each check stops.
- Midjourney AI Detector: What to Check
Can you detect a Midjourney image? Which provenance, metadata and visual clues are worth checking, why detectors miss new versions, and how to ask for the source.
- FLUX AI Image Detector: Evidence Guide
How to review an image suspected of coming from FLUX: PNG generation parameters, IPTC and C2PA metadata, detector coverage limits, and safer next steps.
- Grok AI Image Detector: Evidence Guide
Think an image came from Grok or xAI? Separate the post from the file, check provenance metadata, and learn why attributing an image to Grok is hard.
- Gemini AI Image Detector: SynthID & C2PA
Check if an image or video was made with Google Gemini: use the “Is this made with AI?” SynthID check, read C2PA Content Credentials, and interpret missing signals.
- Stable Diffusion Detector: PNG Info Check
Was an image made with Stable Diffusion? Check PNG Info generation parameters and metadata in your browser, and learn what a missing parameter block means.
- Adobe Firefly Detector: Content Credentials
Was an image made or edited with Adobe Firefly or Photoshop Generative Fill? Check the Content Credentials Adobe attaches, in your browser, and what they cannot prove.
- Meta AI Image Detector: Labels and Metadata
How to check if an image was made with Meta AI: the “Imagined with AI” label, the IPTC metadata and invisible watermarks Meta describes, and what each cannot prove.
Check by AI video generator
- Sora AI Detector: Watermark and C2PA Check
How to tell if a video was made with OpenAI's Sora: the visible Sora watermark, C2PA Content Credentials, what re-encoding and cropping hide, and what to ask next.
- Veo 3 AI Video Detector: SynthID Check
How to check if a video was made with Google Veo: verify SynthID in the Gemini app, look for a visible Veo mark, and read a “not detected” result with care.
A file report is one clue. Review its evidence and limitations.
Check an image