Google has launched a new tool called SynthID Detector, allowing users to check if images, videos, or audio files were generated by AI. The tool scans for imperceptible watermarks embedded by Google's Gemini AI model and other SynthID partners, providing a simple yes or no answer. To use it, users upload a file to SynthID.com and wait for the result. While this is an improvement over previous methods, the tool has limitations. It can't detect AI-generated content from models that don't use SynthID, such as Meta's tools, and it doesn't specify which parts of a file were AI-generated.
Google introduced SynthID in 2023, but the detector was only available to researchers and journalists until now. The tool is designed to prevent false positives by checking for watermarks rather than analyzing media patterns. However, it can't detect AI-generated content from non-SynthID partners, and users are limited to around 10 checks per day to prevent misuse. Additionally, users must be logged in to a Google, OpenAI, or Apple account to use the detector.
SynthID Detector is one of the most reliable AI detection tools available, but it's not the only one. OpenAI, Meta, Anthropic, and Microsoft have also developed their own detection methods. However, none of these tools can reliably spot AI-generated content across the board. Many researchers have created their own AI-spotting tools, but they're not available to the public, and general-purpose AI detectors often produce unreliable results.