All articles tagged with #synthid
Google unveils Gemini 3.1 Flash Live, a faster, more natural-sounding real-time AI audio model for Search, Gemini Live, and developer tools, with SynthID watermarks to flag AI speech. Benchmarks show strong performance in complex tasks, though not always at human levels, and rollout across Google’s products could blur the line between human and bot voices for users.
Google is deploying Lyria 3 in the Gemini app, allowing users to generate 30-second AI music from prompts (and image prompts), with auto lyrics optional, built-in album art, and an embedded SynthID for attribution. It enforces copyright safeguards (not copying named artists), and the feature is live in Gemini web today with mobile rollout in a few days; AI Pro/Ultra subscribers get higher usage limits.
Ring has launched Ring Verify, a public tool that lets users upload Ring videos to see if they’ve been edited, using a C2PA-based metadata signature and automatically included with videos downloaded from December 2025 onward. It signals when a video’s content has been altered (even cropping or length changes), but it can’t confirm real-world authenticity for non-Ring footage from platforms like TikTok or Instagram. Google’s SynthID offers a separate, limited watermark-based check for AI-generated images. While imperfect, Ring Verify represents a step toward provenance and authenticity in an era of convincing AI-generated media.
The Gemini app now allows users to interact with Nano Banana by drawing or annotating on images, and has expanded SynthID's AI video detection to verify if videos were created or edited using Google AI, with features available across Android, iOS, and web platforms.
DeepMind, in collaboration with Google Cloud, has introduced SynthID, a tool for watermarking and identifying AI-generated images. SynthID, currently available in beta for select users of Vertex AI, embeds a digital watermark into the pixels of an image, making it detectable by an algorithm but imperceptible to the human eye. The tool supports Imagen, Google's text-to-image model, and aims to empower users to identify AI-generated content and prevent the spread of misinformation. While SynthID is not foolproof against extreme image manipulations, it provides a promising approach for responsible use of AI-generated content.
Google's DeepMind team has developed a tool called SynthID, which can watermark AI-generated images in a way that is imperceptible to the human eye but easily detectable by AI detection tools. The watermark is embedded in the pixels of the image and remains intact even after cropping or resizing. The tool aims to address concerns about deepfakes and provide a means to identify and detect AI imagery. While SynthID is currently being rolled out for Google Cloud customers, Google hopes to eventually make it an internet-wide standard. The launch of SynthID marks the beginning of an ongoing arms race between AI detection tools and hackers seeking to bypass them.