0
Kakao CI. /Kakao

Kakao is stepping up to build a responsible AI ecosystem in partnership with Google DeepMind, aiming to enhance the transparency and reliability of AI-generated content beyond mere legal requirements.

On May 20, Kakao announced that it signed a technological partnership with Google DeepMind to ensure transparency in AI-generated content. Through this partnership, Kakao will integrate Google DeepMind's "SynthID" into its proprietary AI model, Kanana, establishing a framework to embed and trace watermarks in AI-generated media.

SynthID is a digital watermarking and detection technology developed by Google DeepMind. It embeds imperceptible digital watermarks into images, text, audio, and video, allowing users to verify whether the content was generated by AI without degrading its original quality. Due to its ability to trace AI origins, the technology is drawing significant attention as a tool to counter the misuse of AI, including deepfakes and misinformation.

With this move, Kakao becomes one of the early global adopters of SynthID watermarking and the first Asian enterprise to apply the technology to AI-generated content.

Starting in the second half of this year, Kakao will apply SynthID watermarks to its image generation model, "Kanana Collage," and its video generation model, "Kanana Kinema."

The technology will first be deployed in KakaoTalk's "Kanana Template," a feature powered by the Kanana Kinema model that enables users to easily create short AI videos using images shared within KakaoTalk chatrooms. While videos generated through Kanana Template already display a Kanana logo, the integration of SynthID will ensure that the content can still be identified as AI-generated even if it undergoes editing or secondary processing.

As the AI Framework Act takes effect this year and makes AI content labeling mandatory, Kakao plans to go beyond its statutory obligations by preemptively implementing SynthID’s invisible watermarking technology to bolster transparency and trust in AI-generated content.

"As AI technology rapidly integrates into our daily lives, securing the transparency and credibility of content has become increasingly vital," said Kim Kyoung-hoon, AI Safety Leader at Kakao. "Beginning with the integration of SynthID, we plan to sustain our technical collaboration with Google in the realm of responsible AI, ensuring the delivery of safe AI services that users can trust."

Pushmeet Kohli, Vice President of Research and Scientific/Strategic Initiatives at Google DeepMind, remarked, "Effective collaboration is essential for identifying AI-generated content. By partnering with industry leaders to expand the reach of SynthID technology to more platforms, we will help build a transparent digital ecosystem."