If a photo, video or voice clip looks too wild to be real, you can now check it for free. Google opened its SynthID Detector to everyone on Oct. 7, 2026, at synthid.com. You upload a file, and it tells you whether it carries the hidden watermark Google and a few partner companies put into AI-made images, video and audio. The catch is that it can only find that one watermark, so a clean result does not prove something is real.
The short version
Google’s free SynthID Detector checks images, videos and audio for an invisible AI watermark. It works well for content made with Google’s tools and those of a few partners, and it can’t vouch for anything else.
- The SynthID Detector is free at synthid.com and opened to everyone worldwide on Oct. 7, 2026, in English only.
- It looks for the SynthID watermark, which Google says is in more than 180 billion images and videos and 240,000 years of audio.
- It can spot content from Google’s own AI tools and from partners OpenAI, Nvidia and Kakao, and Google says Apple is coming soon.
- A result of “not detected” does not mean a file is real, because many AI tools add no SynthID watermark at all.
- Use the original file when you can, since screenshots and heavy edits can make the check less reliable.
What the SynthID Detector actually checks
The upshot: this is a watermark reader, not a lie detector. SynthID is a signal Google hides inside AI-made pictures, video and sound when they are created. You can’t see or hear it, but the detector can find it. Google first added the watermark in 2023 and gave an early version of the detector to journalists and other media professionals last year. Now anyone can use it.
According to Google’s announcement, the detector can confirm content made with Google’s AI tools and with tools from OpenAI, Nvidia and Kakao, the South Korean tech company. Google says Apple will be added soon. Google also says its built-in checks in Search, the Gemini app and Chrome already handle more than 1 million requests a day.
How to use the SynthID Detector
- Go to synthid.com on your phone or computer.
- Upload the image, video or audio file you want to check. Use the original file if you can get it, not a screenshot of it.
- Wait for the result. It will say whether a SynthID watermark was found.
- If it was found, treat the file as made or edited with AI from Google or one of its partners.
- If it was not found, keep checking with the steps below before you believe it or share it.
What each result means
| What the detector says | What it means | What to do |
|---|---|---|
| SynthID detected | The file was made or edited with AI from Google or a partner. | Don’t share it as real. If it is being passed around as news, say so. |
| Not detected | No SynthID watermark was found. The file could be real, or made with a tool that adds no watermark. | Check who posted it first, look for the same scene from other angles, and ask the person who sent it where it came from. |
| Detected, but the file looks mostly real | AI was used somewhere, but the result does not say how much. | Treat it as edited. A real photo with an AI touch-up still is not proof of anything. |
What the SynthID Detector can’t do
It can’t catch everything. Many AI image and voice tools, including free open-source ones anyone can run at home, add no SynthID watermark, so their output will come back as not detected. Scammers who clone a voice or fake a photo are not going to pick a tool that labels its own work.
It also works best on clean files. Every time a photo is screenshotted, cropped, compressed or reposted, it becomes a new copy, and heavy editing can weaken a watermark. If a picture reaches you after five WhatsApp forwards, the result is a clue, not a final answer.
When the SynthID Detector is worth a try
- A shocking photo or video of a public figure or a disaster is spreading fast and you want to know before you share it.
- A voice note that sounds like a family member asks for money in a hurry. Check the audio, and then call that person on a number you already have. Our online scam guide covers the other warning signs.
- A listing, review or ad uses a product photo that looks a little too perfect. Our guide to spotting AI fake product reviews pairs well with a quick check.
- You want to know whether a picture your teen found for a school project was made with AI.
Phone makers are working on the other half of the problem, proving a photo is real. Apple says the iPhone 18 Pro can sign camera data at the moment a photo is taken, in an optional Reference mode, so you can later check whether it was changed. Apple also says SynthID support arrives in a software update later this year, as we covered in our iPhone 18 Pro guide. As more cameras and AI tools sign their work, checks like this get more useful.
What to watch
The big one is Apple. When Apple joins, the detector will cover far more of the AI images people actually see, and fewer results will land in the not detected gray zone. Until then, use the SynthID Detector as one quick check among several, and keep asking the oldest question in fact-checking: who sent this, and how do they know?
