[DigitalToday reporter Chi-gyu Hwang] OpenAI has enabled Japanese financial institutions to access its GPT-5.5 model to defend against cyber attacks, Reuters reported on May 29.

According to the report, Japanese Finance Minister Satsuki Katayama (가타야마 사쓰키) told reporters the step was "a major advance in strengthening Japanese financial institutions' ability to defend against cyber attacks" after meeting OpenAI Chief Strategy Officer Jason Kwon in Tokyo.

Earlier, Nikkei reported that Japan's three biggest banks, Mitsubishi UFJ (MUFG) Bank, Sumitomo Mitsui Banking Corp and Mizuho Bank, are expected to be able to access OpenAI's latest models. The models are known to be on a par with Anthropic's Claude Mythos. MUFG and Mizuho declined to comment, while Sumitomo Mitsui did not immediately respond.

Katayama said the Japanese government and financial institutions also plan to secure access to Mythos. Japan launched a public-private joint working group in May to respond to cyber threats that Mythos poses to the financial system.

The emergence of AI models capable of advanced coding has also greatly increased hackers' ability to identify and exploit security vulnerabilities. Against that backdrop, a system in which AI developers provide their latest models first only to trusted partners is emerging as one means of response. OpenAI has previously granted similar access to European companies.