Japan Calls on Infrastructure Operators to Deploy AI Against AI-Powered Attacks
Japan's government is asking systems developers at critical infrastructure sites to use artificial intelligence to detect and patch vulnerabilities before attackers can exploit them. The directive emerged from the first meeting of relevant ministries and agencies on May 18 to address threats from advanced AI models.
Hisashi Matsumoto, the minister in charge of cybersecurity, said cyberattacks using frontier AI models like Claude Mythos-developed by U.S. startup Anthropic-are accelerating in speed and sophistication. Mythos has enhanced capabilities for identifying system weaknesses, which Anthropic restricted from public release due to misuse concerns.
The government expects Tokyo and Japanese financial institutions to gain access to Mythos through cooperation agreements with the United States.
Project YATA-Shield
The government bundled its response into "Project YATA-Shield," a multi-phase approach to infrastructure defense. The plan starts with raising awareness among operators in 15 designated sectors: finance, power, gas, water supply, aviation, railways, ports, and others.
The government will then establish a system to collect and share information on cyberattacks across industries. It will expand public-private partnership programs-previously led by the Financial Services Agency and the Ministry of Economy, Trade and Industry-to cover additional infrastructure sectors.
Officials will also coordinate with foreign governments and AI developers, and work with the Japan AI Safety Institute to provide guidance and update security guidelines as needed.
Sectors Under the Plan
- Finance
- Information and communications
- Government and administrative services
- Health care
- Electric power
- Gas
- Chemicals
- Credit card services
- Petroleum
- Aviation
- Airports
- Railways
- Water supply
- Logistics
- Ports
Matsumoto said the government will "shift into higher gear across the entire government, escalating our efforts not only by one notch but by two or three."
For government officials overseeing infrastructure security, understanding how AI can both defend and threaten systems has become essential. AI for Cybersecurity Analysts covers practical approaches to using AI for threat detection and vulnerability management.
Your membership also unlocks:
