tech
OpenAI gives Japan’s megabanks its newest model for cyber defence
GPT-5.5-Cyber will reach MUFG, SMBC and Mizuho through a verified-defender programme, the finance minister said, as Tokyo treats frontier AI as both threat and shield.
TL;DR
- Japan's MUFG Bank, Sumitomo Mitsui Banking Corporation, and Mizuho Bank will access OpenAI's GPT-5.5-Cyber model.
- The model is provided through OpenAI's "Trusted Access for Cyber" program for vetted defenders.
- The deal involved direct discussions between Japan's Finance Minister Satsuki Katayama and US Treasury Secretary Scott Bessent.
- This initiative is part of Japan's broader push to address AI-related cyber risks, including a public-private working group.
- Anthropic's Claude Mythos is another AI model being considered for similar defensive access by Japanese institutions.
- A potential structural risk is the widening security gap between large, vetted institutions and smaller banks or fintech startups.
- The immediate effect is that three of the world's largest banks will use a frontier AI model for their defences.