Is this image AI? Does every AI picture carry a label or watermark?

Wondering whether a picture you saw was made with AI? Here is what the marks some AI tools add can show, whether they can be removed, how to look for them yourself, and what detection tools get wrong.

Check your own picture

Have the picture? Upload it with the link under the box at the top of this page, and read what the sources say about what it shows and what it claims.

Judge the picture by what it shows and by what the sources say about that, never by an AI score, and look for a hidden label or watermark in the file yourself: one way is further down.

Add the caption or the post that came with the picture: what it says happened needs checking too.

What a label or watermark is

Google DeepMind says: “SynthID embeds digital watermarks directly into AI-generated images, audio, text or video.”

spec.c2pa.org says: “At the heart of the C2PA specification is the Content Credential, a cryptographically bound structure that records an asset's provenance.”

Whether the marks can be removed

Google DeepMind says the watermark doesn't change the image or video quality, is added the moment content is created, and is designed to stand up to modifications like cropping, adding filters, changing frame rates, or lossy compression.

Lifehacker writes: “With enough work, any watermark can technically be removed, though SynthID specifically is pretty resistant to most typical forms of modification.”

How to look for a mark yourself

Google DeepMind says: “Simply upload the image, video or audio clip to your chat, and ask if it's been created or altered by Google AI. Gemini, Search and Chrome will check for a SynthID watermark, and let you know if it finds one.”

Lifehacker puts it plainly: the absence of a watermark cannot prove that an image is authentic.

Finding no mark there leaves the question open, so look at where the picture came from too.

What detection tools get wrong

NewsGuard reported that, according to its findings, the tools collectively declared authentic images to be AI-generated 13.33 percent of the time.

NewsGuard adds that the same set of images did not get the same answer from every tool: “The variation in results — ranging from 27 percent to 93 percent for the same set of images — suggests the tools have different thresholds for what counts as AI manipulation.”

NewsGuard warns: “When detection tools label images with light touch-up edits as fully AI-generated, they risk giving consumers a false basis to dismiss authentic visual evidence.”

news.tuoitre.vn reports on chatbots asked to check pictures: “Internet users are increasingly turning to chatbots to verify images in real time, but the tools often fail, raising questions about their visual debunking capabilities at a time when major tech platforms are scaling back human fact-checking.”

news.tuoitre.vn adds: “In many cases, the tools wrongly identify images as real even when they are generated using the same generative models, further muddying an online information landscape awash with AI-generated fakes.”

news.tuoitre.vn also reports: “Such cases show how AI-generated photos flooding social platforms can look virtually identical to real imagery.”

What to check besides the marks

• Where the picture first appeared, and who posted it first.
• Whether the same scene shows up from another angle, from another photographer or in a news report.
• What the caption says happened, where and when: check the story as well as the picture.
• Then upload the picture to VOM with the link under the box at the top of this page, with its caption.

I already shared the picture. What now?

If you shared it and now doubt it, say so where you shared it, and add what you found. If it turns out to be made with AI, correct the post or take it down.