Tokyo (Jiji Press) — With viral false information on social media becoming a significant societal challenge, Fujitsu Ltd. and the National Institute of Informatics are leading a nationwide effort to develop technologies aimed at addressing the issue.
They have launched full-scale industry-academia collaboration to curb the spread of disinformation, including deepfakes created through generative artificial intelligence.
The term “deepfake” is a portmanteau of “deep learning” and “fake,” describing sophisticated fake images, audio and videos created with the use of AI.
A study by the Australian National University revealed that some of the latest AI-generated facial images are increasingly mistaken for real human faces, with many perceived as more authentic than actual human features. This development makes it nearly impossible for the human eye to discern their authenticity.
Meanwhile, a survey conducted by security software company McAfee LLC in November last year found that 11 pct of Japanese respondents had unknowingly purchased products endorsed by deepfake-generated celebrities.
Junichi Yamagishi, a professor at the NII’s Digital Content and Media Sciences Research Division, expressed concern that “humans tend to be overconfident in their own judgments” and emphasized the need to establish AI-based technologies for assessing authenticity.
In line with this, nine companies and academic institutions, including Fujitsu, the NII and Institute of Science Tokyo, have joined forces to develop the world’s first integrated system to combat false information.
According to a plan announced by Fujitsu in October, the consortium will develop technologies to measure the societal impact of disinformation and to support AI-based authenticity assessments. By integrating these technologies into a comprehensive system, the group aims to mitigate the negative effects of false information and foster a highly reliable digital society. Driven by the belief that this all-Japan initiative will bolster the nation’s economic security, the consortium aims to put the envisioned system into practical use by the end of March 2026.
Leveraging its expertise in detecting fake media, the NII is spearheading the development of technology to identify false information. It also plans to provide an analysis tool capable of estimating falsified parts and the generation method.
“It’s necessary to take a comprehensive approach to determining whether information is false by aggregating a variety of data, not just focusing on deepfake analysis,” Yamagishi said.
Institute of Science Tokyo is tasked with measuring the impact of false information. It plans to visualize users and communities that spread disinformation, providing key insights to assess the societal effects. For its part, Fujitsu is developing a large language model specialized in combating disinformation.
Concerns have also been raised about the so-called backfire effect, a cognitive bias in which individuals reject the classification of specific information as a deepfake when it conflicts with their convictions, ultimately strengthening their belief in the false information.
Kazutoshi Sasahara, a professor at Institute of Science Tokyo, emphasized the importance of devising effective methods for presenting accurate information and optimizing user interfaces. He also highlighted the need for the new system to combat disinformation to incorporate cognitive science research on how users perceive information, alongside developing technologies to verify authenticity and assess the impact of disinformation.