As Google begins to make its newest video-generation instruments obtainable, the corporate says it has a plan to make sure transparency across the origins of its more and more reasonable AI-generated clips. All video made by the corporate’s new Veo mannequin within the VideoFX app can have digital watermarks because of Google’s SynthID system. Moreover, SynthID will be capable to watermark AI-generated textual content that comes from Gemini.
SynthID is Google’s digital watermarking system that began rolling out to AI-generated pictures final 12 months. The tech embeds imperceptible watermarks into AI-made content material in order that AI detection instruments can acknowledge that the content material was generated by AI. Contemplating that Veo, the corporate’s newest video technology mannequin previewed onstage at I/O, can create longer and higher-res clips than what was beforehand attainable, monitoring the supply of such content material will probably be more and more essential.
As generative AI fashions advance, extra corporations have turned to watermarking amid fears that AI may gas a brand new wave of misinformation. Watermarking methods would give platforms like Google a framework for detecting AI-generated content material that will in any other case be inconceivable to tell apart. TikTok and Meta have additionally lately introduced plans to help comparable detection instruments on their platforms and label extra AI content material of their apps.
After all, there are nonetheless vital questions about whether or not digital watermarks on their very own provide adequate safety towards misleading AI content material. Researchers have proven that watermarks may be straightforward to evade. However making AI-made content material detectable indirectly is a vital first step towards transparency.
Compensate for all of the information from Google I/O 2024 proper right here!