Google has introduced AI video verification in its Gemini app, expanding its SynthID watermarking technology beyond images to short videos. The update lets users upload clips up to 90 seconds (under 100MB) and check if any segment was generated or altered by Google’s AI tools, including its Veo video model.
When queried, Gemini scans both audio and visuals for SynthID Google’s invisible watermark that embeds digital signals to verify authenticity. Instead of returning a simple confirmation, the app highlights precise timestamps where AI-generated content is detected, such as “SynthID detected within audio between 10–20 seconds.”
The rollout follows heightened public anxiety over AI-manipulated media on social platforms. Google noted that users are increasingly cautious about sharing posts due to uncertainty over authenticity. Since SynthID’s debut in 2023, more than 20 billion AI-generated images and videos have been watermarked using the technology.
While Google employs SynthID, other AI developers like OpenAI, Meta, and Microsoft rely on the C2PA standard, which embeds metadata to trace content origins. Google said it plans to integrate C2PA into Gemini for broader interoperability across the AI ecosystem.
The new verification tools are now available globally in all languages supported by Gemini, marking another step in the tech industry’s race to enhance transparency and trust amid the surge of generative AI content.
