Google Opens SynthID Detector to Public for Checking AI Watermarks
Google opens SynthID Detector to all users for checking AI watermarks in images, video and audio, with limits.
Google introduced SynthID in 2023 as an imperceptible label embedded in AI-generated content. Until now, ordinary users had no direct way to check for it. Google launched a limited version of the detector in 2025 for certain researchers and journalists and has now expanded access to everyone, Engadget reported.
The detector works by scanning for the watermark rather than analyzing patterns associated with AI generation. Users must log in with a Google, OpenAI or Apple account. A Google spokesperson told Ars Technica that users are limited to about 10 checks per day to prevent them from using the tool's output to find ways to remove SynthID, according to Engadget.
Because it checks for a specific signature, SynthID Detector should avoid false positives that can occur when software misidentifies human-created work as AI-generated. The same design limits its reach. It cannot detect media made by AI models that do not use SynthID, such as Meta's tools, and it does not show which parts of a file were AI-generated.
The tool can scan for SynthID watermarks in content from Google's partners, including OpenAI, NVIDIA and Kakao. Apple also participates in SynthID, but the detector's support for identifying Apple creations is still in the works, Engadget reported. Google says SynthID signatures persist even if AI-generated content is edited after creation.
Other companies offer their own detection tools. OpenAI runs a detector that can identify AI-confirming metadata and SynthID watermarks. Meta embeds a watermark called Content Seal and offers a web-based detector, though a 2026 Reuters report found it sometimes failed to accurately spot them. Anthropic and Microsoft have begun watermarking AI media but do not currently offer consumer tools to detect those marks.
No publicly available detector reliably identifies generative AI across all models. Many researchers have built their own detection tools, but those are not available to the public. General-purpose AI detectors that are openly available produce unreliable results and can flag human-created work as AI content.