Google makes SynthID Detector available globally
AI

Google makes SynthID Detector available globally

TechNews Editorial
TechNews EditorialOct 21, 2026 · 2 min read
Share

Why it matters

SynthID.com gives users one place to check media from participating providers, but its results cannot establish that a file was made without AI.

The facts

  • Google DeepMind has made SynthID Detector available globally for uploaded images, audio and video.
  • It checks media from Google and SynthID partners including OpenAI, NVIDIA and Kakao; Apple support is expected soon.
  • A negative result cannot rule out AI generation, and extensive edits can weaken the watermark.

Advertisement

Google DeepMind has made SynthID Detector available globally. The tool lets users upload images, audio and video to check for SynthID watermarks in media from Google and partners including OpenAI, NVIDIA and Kakao. Support for Apple is expected soon.

Previously, users could ask Gemini whether content was generated by Google AI. Checking media from other providers meant using those providers' separate detection tools. SynthID.com brings checks for participating providers into one place and returns a yes or no result.

A negative result has limits

A negative result does not establish that a file was made without AI. Google says the detector is limited to companies that have adopted SynthID, and a result with “no confident signal” may miss AI-generated content. Media made with Meta's or Microsoft's AI is outside its coverage because those companies use different watermarking technology.

Extensive edits can also weaken a SynthID watermark and produce a false negative. On Reddit, Chicano_Ducky questioned the approach's effectiveness, alleging that “the false positives are unacceptably high on videos their own models made”. Another user, Bulky-Employer, said a low denoising pass could remove a watermark, while calling SynthID “a speed bump against people who want to defraud using AI images”. Those are users' assessments, not findings established by the announcement.

Read nextMicrosoft makes Foundry browser automation generally available

Other methods track a file's history

C2PA metadata offers another way to assess media. It records a file's origin and editing history. Harry G argued on LinkedIn that it is the only “approach that actually scales”, saying that “trying to detect synthetic content after the fact turns into an arms race”. C2PA metadata can be lost when a file is uploaded and does not carry over to screenshots.

Google DeepMind launched SynthID in 2023 to watermark and detect AI-generated images, audio, video and text. It embeds an invisible mark in content using a secret key, with changes to audio frequency probability distributions or to pixel and token patterns. TIME included SynthID in its Best Inventions of 2026 list for its approach to AI-content detection.

Google says it is working to extend SynthID partnerships so the detector can cover more generated media. Apple support is expected soon.

Newsletter

Get the best AI & tech news daily

A concise daily digest. Unsubscribe anytime.

We use your email only to send this newsletter.

Keep reading