Google recently made its SynthID Detector tool available to everyone, letting users check if any image, video, or audio file contains artificial intelligence. The utility scans uploaded media for a concealed marker placed by Google inside synthetic output, delivering an outcome in seconds.

The system functions not only for artificial intelligence output created by Google, but also by OpenAI, Anthropic, Nvidia, ElevenLabs, and others. These organizations employ the identical Synth ID to watermark their synthetic creations.

Here is the process to check files yourself.

How to Check Content for AI Watermarks

  1. Visit the website: Open SynthID.com to access Google's detection tool.
  2. Sign in: Log in using a Google, OpenAI, or Apple account.
  3. Upload your file: Select the image, video, or audio file you want to check.
  4. Check the result: Wait a few seconds for the tool to scan the file and report whether it finds a SynthID watermark.

The outcome assists in establishing if the material bears Google's artificial intelligence watermark. Still, a file lacking this watermark does not automatically mean humans made it, because the utility cannot spot output from every synthetic intelligence platform.

People face a restriction of around 10 checks per day according to reports.

What Can SynthID Detector Find?

Google launched SynthID in 2023 as an invisible watermark placed inside synthetic media. Before the detection utility became accessible to the public, everyday people lacked a direct method to verify if this watermark existed.

Google initially rolled out a restricted edition of the detector for chosen researchers and reporters in 2025. Its public launch now provides everyone a way to check for the watermark without asking Gemini if something came from artificial intelligence.

The system can additionally find SynthID watermarks in synthetic media from Google partners, containing OpenAI, NVIDIA, and Kakao. Apple also takes part in SynthID, though support for spotting Apple-produced material remains under development.

What the Tool Cannot Detect

SynthID Detector searches for a concealed watermark instead of reviewing an image or video for visual indicators of artificial intelligence creation. This method should decrease false positives, but it also restricts what the utility can recognize.

For instance, it cannot spot material produced by Meta artificial intelligence tools because Meta does not use SynthID. Output built by other synthetic systems that omit the watermark can also evade detection.

The utility solely states if it discovers a watermark. It fails to pinpoint the exact sections of an uploaded file created by artificial intelligence.

Google states SynthID watermarks endure alterations applied to synthetic media after its creation. However, users cannot repeatedly test endless versions of a file, since the detector reportedly imposes a daily limit of about 10 checks. A Google representative informed Ars Technica that the limit aims to prevent individuals from repeatedly testing altered files to figure out how to strip the watermark.

Other Ways to Detect AI Content

SynthID is not the sole system built to recognize synthetic media.

OpenAI runs its own detector, which can spot artificial intelligence metadata alongside SynthID watermarks. Meta uses a separate watermark named Content Seal and provides a web-based detector. Nevertheless, a 2026 Reuters report discovered that Meta's utility occasionally failed to correctly recognize synthetic media.

Anthropic and Microsoft have additionally begun watermarking artificial intelligence output, but neither currently supplies consumer-facing utilities for verifying those watermarks.

There remains no single detector that can dependably recognize material created by every synthetic intelligence platform. Most accessible utilities search for specific watermarks or metadata, whereas general-purpose artificial intelligence detectors that inspect media directly can yield unreliable outcomes and sometimes wrongly mark human-made material as synthetic.