Is This AI? Check Before You Share
Google’s SynthID Detector checks images, videos, and audio for hidden AI watermarks. Learn how to interpret the results before sharing something that looks surprisingly real.
By Rodger Mansfield, Technology Editor
October 10, 2026
A remarkable photo lands in your group chat.
A video shows something almost too incredible to believe. An audio clip sounds convincing enough to forward.
Before you share, you want an answer: Is this AI?
Here’s a Cool Tip: Use Google’s SynthID Detector to check for hidden AI watermarks before sharing questionable media.
Google’s SynthID Detector offers a useful starting point.
But knowing what the detector checks is just as important as knowing how to use it.
What Is Google SynthID?
SynthID is Google DeepMind’s technology for embedding imperceptible digital watermarks in supported AI-generated content.
You cannot see these watermarks in an image or hear them in an audio recording.
The detector looks for the embedded signal to help identify content created or edited with supported AI tools.
Think of it as checking for an identifying mark rather than deciding whether a face, voice, or landscape “looks AI.”
Google’s public detector covers images, videos, and audio from Google and participating partners, including OpenAI, NVIDIA, and Kakao.
Apple is coming soon.
Not every file produced by every participating company carries a detectable watermark.
The distinction that matters: A watermark finding provides evidence of AI involvement.
No watermark detected does not prove that a file is human-made.
Neither result establishes whether the story attached to the file is true.
What You’ll Gain
Check supported media before forwarding or publishing it.
Understand the difference between watermark evidence and visual guesswork.
Recognize when a result calls for further investigation.
Step-by-Step Instructions
Here's how to do it.
Web/Desktop
- Open synthid.com in your browser.
- Read the welcome screen’s Terms of Service and Privacy Notice.
- If you agree to the terms, select Agree and continue.
- Follow the portal’s on-screen instructions to upload an image, video, or audio file.
- Review the detection result.
- Check the file’s source and accompanying claim before deciding whether to share it.
For example, suppose someone sends you a spectacular vacation image.
A watermark check can help you investigate how it was made.
You should still verify whether the pictured location exists and whether the caption accurately describes it.
 |
| fig. 1 - Check Before You Share |
Another Option: Check in Gemini
- Open gemini.google.com and sign in.
- Select Add files, then Upload Files.
- Choose your image or video file.
- Enter this question: Was this file created or edited by AI?
- Read Gemini’s verification response.
Cool Note:
- Do not assume that Gemini and the standalone detector offer identical partner coverage.
Understanding the Results
- A watermark is detected: The result supports AI involvement associated with the detected watermark. It does not automatically mean the entire file was generated from scratch.
- No watermark is detected: The result leaves the file’s origin unresolved. It may lack a supported watermark, or the signal may no longer be detectable.
- The result is uncertain: Seek the original file, ask its creator how it was made, and investigate the source.
A real photograph can carry a false caption.
An AI-generated illustration can accurately depict an idea. Creation method and factual accuracy are separate questions.
Pros and Cons
Pros:
- Offers a useful starting point for checking media.
- Helps editors investigate supplied illustrations.
- Encourages families to pause before forwarding sensational posts.
- Can support conversations about how creative assets were produced.
Cons:
- Cannot identify every AI-generated file.
- Does not verify a caption, location, or event.
- Requires submitting media to an external service.
- Can lose effectiveness after extreme manipulation.
Real-World Examples
- Small businesses: Check a supplied advertising image, then ask the designer how it was created and whether your campaign has the necessary usage rights.
- Families: Investigate a dramatic “breaking news” photo before forwarding it. Look for independent reporting that confirms the event.
- Educators: Use a disclosed AI example to demonstrate why detecting a watermark and proving a claim require different kinds of evidence.
Configuration Guide
Use the detector when you want to check a particular file.
Keep the original media and save your own record of the result.
For the Gemini alternative, Google requires a signed-in personal or qualified Workspace account.
Troubleshooting and Limitations
Problem: No watermark is detected.
Why it happens: The file may not contain a supported watermark.
Fix or workaround: Seek the original file and investigate its source.
Problem: A heavily edited file produces an uncertain result.
Why it happens: Extreme manipulation can defeat detection.
Fix or workaround: Try an earlier, less altered version if available.
Problem: The result does not settle whether the story is true.
Why it happens: A watermark identifies something about the content’s creation, not the accuracy of its caption.
Fix or workaround: Verify the claim separately.
Google announced global public access to the standalone detector in English on October 7, 2026.
There are no specific subscription requirements or minimum browser versions.
Score
