Reviewed tool profile

SynthID Detector

Google DeepMind portal that checks images, video and audio for SynthID AI watermarks.

Sources checked Oct 8

Pricing: Google lists no price or paid tier for the portal. Access is open to anyone who signs in, subject to daily usage quotas, and bulk or automated checking is not allowed. Official source 1 ↗ · Official source 2 ↗
Available as: Runs in the browser at synthid.com, hosted by Google; you sign in before uploading, with Google or a third-party account such as Apple. · The same watermark check is built into Google Search, the Gemini app and Chrome, where you upload a file and ask whether Google AI made or altered it. Official source 1 ↗ Official source 1 ↗ · Official source 2 ↗
Primary group: Research & Data

What it does

SynthID Detector is Google DeepMind's web portal for checking whether an image, video or audio file carries a SynthID watermark. SynthID is an invisible watermark added when content is generated by models from Google and its partners OpenAI, NVIDIA and Kakao. You upload a file and the portal reports whether a watermark was found, marking the affected segments for video and audio. It opened to everyone, in English, on 7 October 2026 after a test phase with journalists and media professionals.

Official source 1 ↗ · Official source 2 ↗ · Official source 3 ↗

Features

Image, video and audio checks. Detects a SynthID watermark in images, and marks the segments of a video or audio file that carry one. Official source 1 ↗
Cross-company coverage. Recognises watermarks from generative products by Google, OpenAI, NVIDIA and Kakao; Google says Apple support is coming. Official source 1 ↗ · Official source 2 ↗
Broad file support. Accepts JPG, PNG, WEBP, AVIF, HEIC, TIFF and GIF images; MP4, MOV and WEBM video; and WAV, MP3, OGG, FLAC, AAC and M4A audio. Official source 1 ↗
Plain-language results. Each check returns “watermark detected” or “watermark not detected”, with guidance on what a negative result does and does not prove. Official source 1 ↗
Robust, invisible watermark. The watermark is imperceptible and is designed to survive filters, colour changes and lossy compression. Official source 1 ↗

Pricing & free limits

Public access

No price listedSign-in required; daily usage quotas

Open to anyone worldwide in English. Google publishes no paid tier; usage is rate-limited per account and automated or bulk checking is prohibited. Official source 1 ↗ · Official source 2 ↗

Billing details

Billing details were not established by the checked official sources.

Free access & limits

Each account has daily usage quotas. Creating extra accounts or using automation to get around them breaches the terms, and Google may suspend access. Official source 1 ↗

Best for & limitations

Check whether a viral image, clip or audio recording was generated or edited with Google, OpenAI, NVIDIA or Kakao AI before sharing or reporting on it. Official source 1 ↗ · Official source 2 ↗

Verify the provenance of user-submitted media in a newsroom or fact-checking workflow, alongside other evidence. Official source 1 ↗ · Official source 2 ↗

Not a general AI detector. It only detects media from companies that have adopted SynthID. A “not detected” result does not rule out AI generation, for example by a non-partner model. Official source 1 ↗
Heavy edits can remove the signal. Heavy compression or editing can degrade the watermark, and in rare cases the detector falsely triggers on unwatermarked content. Official source 1 ↗
Generated versus edited is not separated. A detected watermark means the file was likely generated or edited by a SynthID-enabled model; the result does not say which. Official source 1 ↗
No bulk or API use. The terms forbid submitting requests programmatically or running bulk verification through the portal. Official source 1 ↗
English only. The portal launched globally with an English-only interface. Official source 1 ↗

SynthID Detector product screenshot — Website previewFixed desktop viewport · 1440 × 900