Google opens SynthID Detector to users globally

Google said it was expanding access to everyone, with the detector available globally in English starting the day of its announcement.[1] Anyone can use the tool at synthid.com by logging in.[2] Google introduced an early version the previous year to help media professionals verify AI-generated content.[1] The expansion makes the tool available to a broader group, but the evidence does not describe a paid tier, upload limit, or eligibility rules beyond logging in. For teams evaluating it, the practical first step is to test whether their file types and source media fit the tool’s scope.

Google says SynthID Detector checks image, video, and audio files for content made with AI from Google or its partners.[1] Google named OpenAI, NVIDIA, and Kakao as partners whose AI-generated content it can check, and said Apple would be added soon.[1] That means the tool is not described as a universal AI detector: it looks for SynthID, a watermarking technology Google launched in 2023 for identifying AI-generated media.[1][3] A team considering the tool should first ask whether the media source uses SynthID, rather than treating the detector as a complete authenticity check.

Because the detector identifies media from companies that have adopted SynthID, a file with no detected watermark may still have been created by a model that does not use it or may have been heavily modified.[3] Google also says the detector can rarely report a watermark on content that does not carry one.[3] Google recommends uploading the highest-quality file available and using multiple points of data before drawing a conclusion.[3] The result should therefore inform verification, not replace it.

Google says it launched SynthID in 2023 and has watermarked more than 180 billion images and videos with the technology.[1] It also says SynthID has been used to watermark 240,000 years of audio content.[1] The public detector joins verification features in Search, the Gemini app, and Chrome, which Google says now regularly handle more than one million requests per day.[1] Those figures describe Google’s reported use of its watermarking and verification features; they do not establish how often the new public detector will correctly identify AI-generated media from other sources.