Generative AI information leakage risk: Concerns among Japanese companies stand out; more than 40% of frontline employees say, “We would like to use it if permitted for work”

According to the 2026 edition of a Ministry of Internal Affairs and Communications survey, Japanese companies cited security risks, including information leaks associated with the use of generative AI, as their greatest concern, at 46.4%. The figures were 42.7% for China, 36.9% for the United States and 31.7% for Germany, showing that concern was particularly pronounced in Japan.

Copyright infringement ranked highest among concerns in Germany, while accuracy of output ranked highest in the United States and China. Security risks did not rank among the leading concerns in countries other than Japan. Japanese companies showed a relatively lower level of concern in areas other than security.

The particular caution Japanese companies show toward information-leak risks reflects the possibility that AI may use input data for training, as well as the risk of theft through unauthorised access by third parties. Many companies still in principle prohibit the use of generative AI internally, and introducing it will require security measures and the development of usage guidelines. The survey also found that 43.2% intended to use generative AI “when its use for business purposes is approved internally,” while 82.4% of those already using it had experienced its benefits.