SoftBank launches OpenAI-powered cybersecurity service for Japan's top companies

SoftBank Group rolled out a new cybersecurity service built on OpenAI technology, aiming it at Japan’s top 3,000 companies, including operators linked to airports, power systems and transportation. Announced in Tokyo on June 16, the move pushes generative AI further into the role of critical defense infrastructure, not just a tool for writing code or text.

Masayoshi Son, SoftBank’s chairman and chief executive, cast the threat in stark terms, calling Japan’s exposure to cyberattacks “a crisis.” He said attackers are increasingly using AI themselves, and compared the shift in danger to moving from rifle shots to machine guns. SoftBank said its new product is called “Patching as a Service,” and Son said it is designed to diagnose weaknesses first and then help customers close them before criminals can exploit them.

The company said it had already tested the approach inside SoftBank Corp., where its cybersecurity team ran a large-scale internal vulnerability assessment using OpenAI’s cybersecurity technology. SoftBank said that effort turned up promising results and that the lessons from it will be used in customer deployments. The launch gives SoftBank a concrete product to attach to a partnership that has been taking shape for months, and it signals that the company wants to sell AI not only as a productivity boost, but as a tool for defending the systems that keep the country running.

OpenAI chief research officer Mark Chen appeared at the Tokyo event in place of chief executive Sam Altman, who could not attend because his baby daughter arrived earlier than expected. The two companies have been tightening their ties in Japan through SB OAI Japan GK, a 50-50 joint venture launched on November 5, 2025, to provide “Crystal intelligence” for major Japanese enterprises. SoftBank said the broader offering will be rolled out through that venture, and that SoftBank Corp. will be the first user of it.

For Japan’s biggest companies, the service is more than a software launch. It is a test of whether the same AI systems that expand the attack surface can also be trusted to guard it, especially in sectors where a breach could ripple far beyond a single balance sheet and into public safety.