Japan’s AI Safety Institute has assessed that generative AI is already contributing to both an increase in the volume of cyberattacks and an improvement in their quality, prompting the country’s digital minister to call for urgent preventive measures.

Digital minister Toshiharu Furukawa urged individuals and organisations to take action at a news conference on October 6, following a sharp surge in corporate cyberattacks driven by malicious AI tools. He said prevention was not just a problem for companies and organisations, advising people to avoid reusing passwords, use multi-factor authentication, and treat suspicious emails and social media messages with caution.

A sharp rise in attacks

The National Police Agency recorded 611 cyber incidents involving personal-data leaks by early October 2026, while suspicious IP-address access attempts reached 13,687 per day during the first half of the year, roughly 50 percent higher than the same period in 2025.

Recent breaches illustrate the scale of the damage. Discount retailer Mr Max Holdings disclosed that unauthorised access may have affected up to 1.73 million customers, while Monogatari Corporation reported the leakage of personal information concerning about 10.78 million members. Asahi Kasei Therapeutics indicated possible exposure affecting roughly 510,000 people, and GMO Research and AI reported the loss of up to 950,000 records along with fraudulent use of customer reward points.

On October 7, IDC Frontier, a SoftBank subsidiary, confirmed a ransomware attack had disrupted its IDCF Cloud platform, affecting its East Japan Region 1 and disrupting services used by 495 companies and local governments.

Government responds

The government is moving towards a more preventive and coordinated approach rather than relying only on incident response. In June 2026, the Digital Agency revised its guidance for the procurement and use of generative AI across government, extending governance requirements to areas such as image and audio generation and bringing AI agents within the scope of reporting to its advanced-AI advisory structure.

The AI Safety Institute’s assessment highlights a dual reality: the same technology driving rapid innovation is also handing attackers faster, more effective tools, making trustworthy evaluation of AI systems an urgent national priority.