Google's AI Detector Goes Global, but It Only Catches Watermarked Media

Google made its SynthID Detector available to anyone on 7 October 2026, expanding a tool that had been limited to select journalists, media professionals, and researchers since testing at Google I/O. The detector lets users check whether an image, video, or audio file was made with AI from Google or its partners, a list that includes OpenAI, NVIDIA, and Kakao, with Apple support described as coming soon.

It joins SynthID verification features already built into Search, Gemini, and Chrome, which Google says handle more than 1 million requests daily. Since launching SynthID in 2023, Google says it has watermarked over 180 billion images and videos and 240,000 years of audio.

That scale is real, and a free public detector is a genuine step for provenance. But the mechanism defines the limit. SynthID detection works by reading watermarks that Google and its partners embed at generation. Media produced by tools that do not embed SynthID — the majority of what circulates — will not be flagged, because there is nothing to read.

The gap matters because AI-generated misinformation is already being monetised at scale. BBC Verify documented AI-generated videos and fabricated satellite imagery about the Iran war amassing hundreds of millions of views. Separately, EU rules have required labels on authentic-looking AI-generated content since 2 August 2026. A detector that only reads partner watermarks cannot enforce that regime on its own, and it cannot reach the content driving the traffic.

The honest reading: this is a milestone in AI-content provenance, not a solution to AI misinformation. It verifies what Google and a few partners choose to mark. Everything else stays unverified — and the detector will not say so.

- https://blog.google/innovation-and-ai/models-and-research/google-deepmind/synth-id-ai-content/ - https://techcrunch.com/2026/10/07/googles-new-synthid-website-can-identify-ai-generated-media/ - https://www.bbc.com/news/articles/ckg8wvz427vo - https://www.theguardian.com/technology/2026/jul/31/ai-labels-to-be-compulsory-on-authentic-looking-content-under-eu-rules