Google Opens SynthID AI Detector to the Public Worldwide

Key takeaways

• Google launched its web-based SynthID Detector publicly at synthid.com in English after a year of restricted testing with media professionals.
• The tool identifies invisible SynthID watermarks across supported image, video, and audio formats generated by Google, OpenAI, Nvidia, and Kakao.
• Users must log in with a Google, OpenAI, or Apple account; Apple Intelligence support is scheduled to roll out soon.
• SynthID Detector is not a universal checker and cannot identify AI media from unsupported systems like Anthropic, xAI, or Meta.
• The detector flags whether content was made or edited with supported tools, but cannot isolate which specific parts of a file were modified.

On October 7, 2026, Google officially expanded access to its SynthID detector tool by launching a dedicated public website at synthid.com. First introduced as an imperceptible watermarking system in 2023, SynthID embeds invisible signals directly into generative files during creation. Google initially restricted early versions of the detector tool to journalists, researchers, and media professionals following testing at Google I/O in 2025. According to an announcement from Google's official blog, the standalone detection tool is now available globally in English, allowing anyone to inspect digital files for provenance signatures.

The public release allows users to check digital media created across several major AI platforms. Beyond Google's internal generation models, the web tool currently identifies SynthID watermarks embedded by systems from OpenAI, Nvidia, and Kakao, with integration for Apple's generative tools slated to arrive in the near future.

Supported File Formats and Partner Integrations

The web tool checks a broad selection of multimedia formats, as detailed by TechCrunch. For still graphics, SynthID Detector supports JPG, JPEG, PNG, BMP, WEBP, AVIF, HEIC, HEIF, TIFF, TIF, and GIF files. Video uploads accept MP4, MOV, and WEBM formats, while audio compatibility includes WAV, MP3, OGG, FLAC, AAC, and M4A.

Access to the tool is free, though Engadget reports that users must authenticate using an existing Google, OpenAI, or Apple account before uploading media. The site prompts visitors to supply the highest quality version available to improve detection accuracy.

Google models utilizing SynthID include its Nano Banana, Veo, and Lyria systems, as well as generative interfaces across Gemini, Flow, ProducerAI, and Google Vids. In its release announcement, Google stated that more than 180 billion images and videos alongside 240,000 years of audio content have been marked with SynthID watermarks since 2023. Dedicated verification capabilities are also integrated into Google Search, the Gemini mobile app, Circle to Search, and Google Chrome, where Google says verification volume exceeds 1 million daily requests.

Apple is anticipated to be the next major platform integrated into the tool. As reported by 9to5Google, images generated or edited via Apple Intelligence and features such as Image Playground will incorporate SynthID watermarks in an upcoming update.

Detection Limitations and Watermarking Blind Spots

While the web detector broadens access to provenance checks, both Google and industry analysts emphasize that it is not an all-purpose AI scanner. As highlighted by The Decoder, the system flags only media embedded with a valid SynthID watermark and cannot verify synthetic content produced by platforms outside the partnership.

Because rival AI vendors deploy proprietary techniques or no watermarks at all, large segments of synthetic content remain undetectable by SynthID. As SiliconANGLE pointed out, Anthropic's Claude, xAI's Grok, open-weights Chinese models, and Meta's standalone generation tools do not adopt SynthID. Meta, for instance, has tested its own detection systems built specifically for its Muse Image models, leaving the industry split across fragmented mechanisms.

Another technical limitation lies in how detections are reported. As explained by Search Engine Journal, when SynthID detects a watermark, its readout indicates that the file was "made or edited" using supported AI tools and warns that the asset may have undergone further edits. The tool does not isolate which specific pixels, frames, or audio frequencies were modified, meaning an image touched up with generative editing generates the same high-level warning as a fully synthetic output.

Google's documentation further notes that watermarks can sometimes be stripped during processing and that no single detection signal is completely robust, advising users to evaluate multiple contextual signals before drawing conclusions.

Frequently asked questions

Where can users access the SynthID Detector?

The detector is available at synthid.com and is free to use with a Google, OpenAI, or Apple account login.

Which companies currently support SynthID watermarking?

Media generated by Google, OpenAI, Nvidia, and Kakao can be detected, with Apple Intelligence support scheduled to follow soon. ElevenLabs previously announced commitments to integrate the watermark.

Can SynthID Detector detect all AI-generated images and audio?

No. The tool only identifies media that contains an intact SynthID watermark. It cannot detect AI-generated content produced by non-participating models, such as Anthropic's Claude, xAI's Grok, or Meta models.

Does SynthID show which parts of an image were edited?

No. The tool indicates whether an asset was made or edited with supported AI, but it cannot differentiate between fully synthetic media and media that received light AI edits.

Sources

1. We're making it easier to identify AI-generated content globally.Google · Oct 7, 2026 · Official
2. Google’s new SynthID website can identify AI-generated mediaTechCrunch · Oct 7, 2026
3. Google's AI Detection Website Is Now AvailableEngadget · Oct 7, 2026
4. Google’s SynthID AI content detector expands globally, will work on Apple’s models ‘soon’9to5Google · Oct 7, 2026
5. Google says 180 billion images and videos now carry SynthID watermarks as detector goes publicThe Decoder · Oct 7, 2026
6. Google debuts SynthID Detector tool for flagging AI-generated content, but it’s far from perfectSiliconANGLE · Oct 8, 2026
7. Google Opens Its SynthID Watermark Detector To The PublicSearch Engine Journal · Oct 7, 2026

How this story was made: the newsroom picked it up from techcrunch.com, Techmeme and engadget.com, gathered the full text of the sources above, and drafted it with AI assistance. Every factual claim was then checked against those sources before publishing (24 claims checked). Illustrations marked as AI-generated are not photographs. Spotted an error? Tell us.

#Google #SynthID #Generative AI #AI Detection #Watermarking

Published October 9, 2026 at 02:06 UTC