OpenAI has given some Japanese financial institutions access to its latest GPT-5.5 model, as banks prepare for a new wave of cybersecurity risks created by more powerful artificial intelligence systems.
Japanese Finance Minister Satsuki Katayama said on Friday that OpenAI gave some Japanese financial institutions access to GPT-5.5 to help prevent cyberattacks.
Japan sees early AI access as cyber defense
The move comes as governments and banks worry that advanced AI models could be used by hackers to find and exploit software weaknesses faster than before.
Reuters reported that new AI models able to code at a high level have given hackers an unprecedented ability to identify cybersecurity risks and create ways to exploit them.
The Straits Times also reported that early access to new models available only to trusted partners is one possible defense for banks and other companies at risk of being destabilized.
Katayama did not name the specific institutions receiving access. However, Reuters quoted her as saying the access was “a big step forward” in strengthening Japanese financial institutions’ ability to defend against cyberattacks.
Japan’s biggest banks may be involved
The announcement follows reports that Japan’s largest banks could receive access to OpenAI’s newest model.
Japan’s three biggest banks — MUFG Bank, Sumitomo Mitsui Banking Corp, and Mizuho Bank — were expected to gain access to OpenAI’s latest model. The model is believed to be on par with rival Anthropic’s Claude Mythos.
MUFG and Mizuho declined to comment, while Sumitomo Mitsui Banking was not immediately available to comment. The banks have not publicly confirmed whether they are among the institutions receiving access.
Mythos also part of Japan’s cybersecurity plan
Japan is also preparing to use Anthropic’s model for financial cybersecurity.
Katayama said Japan’s government and financial institutions were also expected to gain access to Mythos to enhance their defenses. Japan set up a public-private working group in May to address cybersecurity risks to the financial system posed by Mythos.
That detail matters because the same AI systems that can help attackers may also help defenders. Banks could use advanced models to examine code, detect vulnerabilities, test security assumptions, analyze threat patterns, and respond faster to emerging cyber risks.
OpenAI access follows government talks
The decision appears to have followed government-level negotiations.
OpenAI’s decision to grant Japanese financial institutions access came after negotiations between the Japanese and U.S. governments. The move followed similar access given to European companies.
The access also shows how frontier AI models are becoming part of national cybersecurity strategy. Banks are critical infrastructure, and governments are increasingly treating AI capability as both a potential threat and a defensive tool.
AI becomes a banking security priority
For Japan, the OpenAI deal gives financial institutions early access to tools that may help them prepare for AI-powered cyberattacks. For OpenAI, the move places GPT-5.5 inside one of the world’s most important financial systems under a trusted-access model.
The larger issue is that AI is changing the speed of cybersecurity. If attackers can use advanced models to find flaws faster, banks may need equally advanced systems to defend themselves. Japan’s move suggests that early access to frontier AI models may soon become part of how governments protect banks, markets, and financial infrastructure.
