Arya News - The agency aims to prevent serious incidents before they occur, by utilizing AI to quickly identify social media posts that pose a high risk of leading to crime and issuing warnings.
TOKYO – The National Police Agency plans to use generative AI to strengthen measures against “lone offenders,” individuals who become radicalized without belonging to any specific organization, according to sources.
The agency aims to prevent serious incidents before they occur, by utilizing AI to quickly identify social media posts that pose a high risk of leading to crime and issuing warnings.
In 2022, former Prime Minister Shinzo Abe was fatally shot while giving a speech in Nara City. As part of efforts to strengthen its countermeasures in response to this crime, the NPA established an internal section dedicated to addressing lone offenders in April 2025.
Tasked with compiling information on early warning signs, the section has been manually analyzing social media data and issuing warnings to users whose posts hint at harm toward prominent people and others.
The NPA plans to utilize AI to streamline these efforts and has already conducted pilot tests. It envisions a system that identifies posts containing key words that could be interpreted as death threats and then assesses the level of risk based on the context of preceding and following posts.
In addition, as part of “active cyber defense” aimed at preventing serious cyberattacks before they occur, the agency plans to utilize AI-powered equipment to infiltrate the servers from which attacks originate and neutralize them.
On Friday, the NPA included a total of approximately ¥8.98 billion in its preliminary budget request for next fiscal year to cover these AI-related expenses. Separately, it plans to utilize cross-ministerial budgets to develop technology that uses AI to assess the urgency of emergency calls and inquiries from the public.