Google Gemini Fights Deepfakes: Detects Hidden AI Watermarks in Images, Videos

• AI-generated media is rapidly spreading, making it challenging to distinguish real content from manipulated visuals.
• Google's Gemini tool now helps users verify images and videos by detecting hidden AI watermarks called SynthID.
• Users can upload media (videos up to 100MB/90 seconds) to the Gemini app and prompt it to check for AI generation.
• Gemini scans for SynthID, which is embedded in both audio and visual layers, identifying AI-generated elements.
• SynthID watermarks are invisible to humans, detectable only by AI systems, and difficult to remove without degradation.

Why It Matters: Gemini uses invisible SynthID watermarks to verify AI-generated content, enhancing media authenticity.