KaKao

- Kakao adopts Google Deepmind’s digital watermarking and detection technology  SynthID applying it to its proprietary AI model ‘Kanana’
- Beginning in the second half of this year, invisible watermarks will be embedded into AI services based on image and video models to enhance transparency and trustworthiness in AI-generated content
- Expanding collaboration in Responsible AI to strengthen the transparency and reliability of AI-generated content

 

[2026-05-20] Kakao is accelerating efforts to build a responsible AI ecosystem in collaboration with Google DeepMind.

 

Kakao (CEO Shina Chung) announced on the 20th of May that it has entered into a technology partnership with Google DeepMind to help create transparency in AI-generated content. Through this partnership, Kakao will apply ‘SynthID’ to its proprietary AI model ‘Kanana,’ establishing a technological foundation for watermarking and detecting AI-generated content.

 

Kakao is one of the early adopters of SynthID watermarking technology globally and the first in Asia to integrate SynthID watermarking into its AI-generated content. 

SynthID is a digital watermarking and detection technology developed by Google DeepMind. It embeds digital watermarks into images, text, audio and video that are imperceptible to the human eye and ear, enabling AI-generated outputs to be verified. In particular, the technology has gained attention as a countermeasure against the misuse of AI-generated content such as deepfakes and misinformation, as it can identify whether content was AI-generated without compromising content quality.

 

Beginning in the second half of this year, Kakao will apply SynthID watermarks to its image model ‘Kanana-Kollage’ and video model ‘Kanana-Kinema.’

 

Kakao first plans to apply SynthID watermarking to the KakaoTalk feature ‘Kanana Template.’ Kanana Template is a feature that allows users to easily create short AI-generated videos using images exchanged on KakaoTalk, powered by the ‘Kanana-Kinema’ model. Videos generated through Kanana Template now display the Kanana logo by default, while SynthID will additionally enable identification of AI-generated content even if the output is later edited or otherwise modified.

 

With the enforcement of the AI Basic Act this year and the full-scale introduction of the AI-generated content labeling system, Kakao is proactively going beyond legal obligations by applying invisible watermarking technology based on SynthID to strengthen the transparency and reliability of AI-generated content.

 

Kim Kyunghoon, AI Safety Leader at Kakao, said, “As AI technology rapidly becomes integrated into everyday life, the importance of securing transparency and trustworthiness in AI-generated content is also growing.” He added, “Starting with the adoption of SynthID, we will continue technological collaboration with Google in the area of Responsible AI and provide AI services that users can trust and use with confidence.” 

 

Pushmeet Kohli, VP, Science and Strategic Initiatives, Google DeepMind mentioned "Helping people identify AI-generated content requires effective collaboration. By partnering with industry leaders to bring our SynthID technology to more platforms, we are taking a crucial step toward building a more transparent digital ecosystem for all." [End]