Skip to content
Today
Sections

Product Reports disagree19 publishers 2 min read Published Updated

SynthID.com lets anyone test a file for watermarks from Google, OpenAI, Nvidia and Kakao

Google opened SynthID.com to anyone, checking images, video and audio for OpenAI, Nvidia and Kakao watermarks as well as its own. That speeds up flagging partner-made media but cannot vouch for files from outside the SynthID group.

The Product Desk

Send a correction

What happened

  • Before the launch, anyone outside Google's trusted tester program could check a file for SynthID only by asking Gemini.
  • Select journalists, media professionals and researchers had been testing the tool since Google I/O last year, and Google opened it to everyone on Tuesday.
  • The public site returns only a message saying whether SynthID was used to produce or edit a file, without the internal tool's highlighting of watermarked regions.
  • It accepts 20 listed file types across images, video and audio, including HEIC and AVIF photos, MOV video and FLAC audio.
  • Apple is due to add SynthID support soon, according to Google.

Why it matters

  • What's now possible capability A reviewer no longer needs to know which vendor made a file before picking a detector, because one upload checks all the partners' watermarks.
  • What it costs cost One message covers both generated and edited media, so anyone who needs to know which part of a photo was changed still has to find that out another way.
  • Who's exposed exposure Anyone passing off partner-made images as real now faces a check that any recipient can run on a public website, with no need to go through Gemini.

Take a ChatGPT image that a reviewer ran through Google's old detector. The file carried SynthID, and the detector still would not flag it, because it read only Google's own version of the watermark [9]. Asking Gemini instead had its own problem. Ars Technica found the model's answers could be confusing to someone who just wanted a yes or no [20].

In TechCrunch's summary, the pitch is a site that lets anyone verify whether a piece of media is AI-generated [6]. The site checks for SynthID, an invisible signal written into the pixels of images and video and into the waveform of audio [4]. Google stamps it on output from its Nano Banana, Veo and Lyria models and from tools including Gemini, Flow, ProducerAI and Vids [16]. By Google's count, Gemini alone has put it into more than 180 billion images and videos and 240,000 years of audio [3].

A negative result covers much less ground. Microsoft and Meta have their own watermarking standards [10], so output from their models is outside what SynthID.com looks for. OpenAI, a SynthID partner, also keeps its own checking site [1][8]. TechCrunch noted that these tools are not infallible and often fail to identify content made by their makers' own models [11].

Google has built SynthID checks into the Gemini app and Chrome, and says people make 1 million verification requests a day [17]. Neither report includes a figure for how often those checks are right.

For a team adding this to a review queue on Monday, the useful grid has two axes. One is whether the decision needs to show a file is AI-made or show that it is real. The other is whether the site said yes or no. A yes on a file you suspect is AI-made is a flag you can act on, as long as the file came from a partner's model. A no on that same file settles little. If the decision is about clearing a file as real, the site can only rule a file out, and only with a yes.

I'd put SynthID.com at the front of any queue that flags content as AI-made and keep it out of any step that approves content as authentic, since a no from the site leaves the reviewer with the same work they had before the upload.

What to watch

  • Whether Microsoft or Meta join SynthID, or Google adds their watermark standards to the site, which would widen what a negative result covers.
  • Whether Apple's SynthID support ships and its watermark starts registering in SynthID.com checks.
  • Whether OpenAI keeps its separate checking site now that Google's public detector reads its watermark too.

Clarity's read

What the record supports and how the coverage leans. The claims behind it follow.

Reality

Evidence72
Adoption40
Hype gap+15
Incentives55
Confidence68

Perspective Coverage

20 publishers
Builder
Builder 27%
Operator
Operator 54%
Investor
Investor 19%
Why these scores

Claim ledger

Ranked by verification strength, evidence, and original report placement.

  1. [1]

    Google says the expanded detector supports the distinct watermarks of its SynthID partners, including OpenAI, Nvidia and Kakao.

    Reported Supported Source: Google, via Ars TechnicaView cited source
  2. [2]

    Google announced that its SynthID detector now supports watermarks from all its partners and is available for anyone to use on a new dedicated website, SynthID.com.

    Reported Supported Source: Ars Technica, reporting Google's announcementView cited source
  3. [3]

    Google says Gemini alone has integrated SynthID with over 180 billion images and videos, plus 240,000 years of audio.

    Reported Supported Source: Google, via Ars TechnicaView cited source
  4. [4]

    SynthID is an invisible signal encoded in the pixels of AI-generated images and videos and in the waveform of AI audio.

    Reported Supported Source: Ars TechnicaView cited source
  5. [5]

    Google launched the site on Tuesday, having made the tool available to select journalists, media professionals and researchers for testing at Google I/O last year.

    Reported Supported Source: TechCrunchView cited source
  6. [6]

    TechCrunch described the site as letting anyone verify whether a piece of media, be it an image, a video or an audio clip, is generated using AI.

    Reported Supported Source: TechCrunchView cited source
  7. [7]

    Google says Apple will soon be among the SynthID partners whose watermarks the detector supports; TechCrunch reported Apple is said to be adding support soon.

    Reported Supported Source: Google, via Ars Technica; TechCrunchView cited source
  8. [8]

    OpenAI maintains its own site for checking content.

    Reported Supported Source: TechCrunchView cited source
  9. [9]

    Google's previous detector only checked for Google's version of SynthID; ChatGPT images, which carry SynthID, would not be flagged by it.

    Reported Supported Source: Ars TechnicaView cited source
  10. [10]

    Microsoft and Meta have their own standards for watermarking and verifying content.

    Reported Supported Source: TechCrunchView cited source
  11. [11]

    TechCrunch reported that these tools are not infallible, often failing to identify content created by their makers' models.

    Reported Supported Source: TechCrunchView cited source
  12. [12]

    Users can check a file for the involvement of all the partner AI systems on the site instead of accessing each partner's detector individually.

    Reported Supported Source: Ars TechnicaView cited source
  13. [13]

    The site supports JPG, JPEG, PNG, BMP, WEBP, AVIF, HEIC, HEIF, TIFF, TIF and GIF image formats (11 listed).

    Reported Supported Source: TechCrunchView cited source
  14. [14]

    The site supports MP4, MOV and WEBM video formats (3 listed).

    Reported Supported Source: TechCrunchView cited source
  15. [15]

    The site supports WAV, MP3, OGG, FLAC, AAC and M4A audio formats (6 listed).

    Reported Supported Source: TechCrunchView cited source
  16. [16]

    Google's Nano Banana, Veo and Lyria models, and tools with generation capabilities including Gemini, Flow, ProducerAI and Vids, use SynthID to watermark media they generate.

    Reported Supported Source: TechCrunchView cited source
  17. [17]

    Google has built SynthID verification into the Gemini app and Google Chrome, and the company said people currently make 1 million requests to verify content every day.

    Reported Supported Source: Google, via TechCrunchView cited source
  18. [18]

    The site accepts 20 listed file types across images, video and audio.

    Derived Supported
  19. [19]

    Until now, the only way for those outside Google's trusted tester program to check for SynthID was to ask Gemini.

    Reported Contested Source: Ars TechnicaView cited source
  20. [20]

    Ars Technica described checking via Gemini as clunky, saying the model's output could sometimes be confusing when a user just wanted a simple yes or no.

    Reported Contested Source: Ars TechnicaView cited source
  21. [21]

    The website gives less information than the internal tool, which highlighted regions of an image showing SynthID pixels; it returns only a message saying whether SynthID was used to produce or edit the media.

    Reported Contested Source: Ars TechnicaView cited source

Sources

19 independent publishers whose own reporting we read for this story.

  1. 9to5google.com

    2 articles · October 7, 2026

    Google's SynthID AI detector will work with Apple's models 'soon'
  2. arstechnica.com

    2 articles · October 7, 2026

    Google rolls out improved SynthID AI content detector, now available globally
  3. cnet.com

    1 article · October 8, 2026

    Is That Picture AI? Google’s SynthID Detector Launches to Identify AI-Generated Media - CNET
  4. cryptobriefing.com

    1 article · October 7, 2026

    Google opens its SynthID detector to everyone, adds partner watermarks
  5. cryptopolitan.com

    1 article · October 7, 2026

    Anyone can now check files against Google's SynthID AI watermark, applied to 180 billion images and videos
  6. dev.to

    1 article · October 9, 2026

    Google Opens SynthID Detector Globally in English for AI Media Verification
  7. digitaltrends.com

    1 article · October 7, 2026

    Google’s SynthID Detector just launched globally to help you spot AI-generated media - Digital Trends
  8. engadget.com

    2 articles · October 7, 2026

    Google's AI detection website is now available
  9. gizmodo.com

    1 article · October 7, 2026

    Google Launches New Website to Detect AI Fakes, But Don’t Assume It’s Perfect
  10. gsmarena.com

    1 article · October 8, 2026

    Everyone can now check if an image, video, or audio file was made with AI - GSMArena.com news
  11. macrumors.com

    2 articles · October 7, 2026

    Google's SynthID AI Detector is Now Available to Everyone - MacRumors
  12. pcmag.com

    1 article · October 7, 2026

    Is That Image, Audio, or Video Made by AI? Google's AI Detector Will Tell You | PCMag
  13. searchenginejournal.com

    1 article · October 7, 2026

    Google Opens Its SynthID Watermark Detector To The Public
  14. siliconangle.com

    2 articles · October 7, 2026

    Google debuts SynthID Detector tool for flagging AI-generated content, but it's far from perfect - SiliconANGLE
  15. techcrunch.com

    2 articles · October 7, 2026

    Google launches SynthID Detector, a website that lets users detect AI-generated image, video, and audio media across dozens of common file formats
  16. techrepublic.com

    1 article · October 8, 2026

    Google’s SynthID Detector Is Now Public: Here’s What It Can Actually Detect
  17. thenextweb.com

    1 article · October 7, 2026

    Google opens SynthID Detector to everyone to check for AI-made media
  18. theverge.com

    1 article · October 7, 2026

    Google’s AI content detector is rolling out globally. | The Verge
  19. thurrott.com

    1 article · October 8, 2026

    Google Makes its SynthID AI Content Detector Available to Everyone
  20. uk.pcmag.com

    1 article · October 7, 2026

    Is That Image, Audio, or Video Made by AI? Google's AI Detector Will Tell You

Topics and entities

Topics

Entities