SoftBank and OpenAI are moving into cyber defence. The two said on Tuesday they are launching “Patching as a Service,” a security product built on OpenAI’s technology, to shield the companies behind Japan’s critical infrastructure from a rising wave of cyberattacks.
SoftBank founder Masayoshi Son framed the threat in stark terms, calling Japan’s exposure “a crisis” and likening today’s AI-powered attacks to machine guns replacing the rifle shots of the past.
What the service actually does is narrower than the name suggests. It runs an AI-driven vulnerability assessment, then helps plan how to fix the weaknesses it finds and advises on carrying the work out. It stops short of applying the patches itself, and SoftBank is clear that expert human teams still do the prioritising and the planning.
The first customers will be operators of national infrastructure: at the launch Son put the target at around 3,000 firms behind Japan’s airports, power systems and transport, though the official release describes a more measured, progressive outreach to “selected eligible companies.”
Why SoftBank and OpenAI are selling security
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!
The pitch rests on a simple, uncomfortable logic: attackers are already using AI to automate and scale their attacks, so defences have to become just as AI-savvy. The same models that probe systems for flaws can be turned around to find them first.
SoftBank says it tested exactly that on itself, running a large-scale assessment of its own internal systems with OpenAI’s cyber models and reporting “promising results,” then carrying those lessons into the product.
The launch runs through SB OAI Japan GK, the 50:50 joint venture SoftBank and OpenAI set up last year to develop and exclusively sell OpenAI-based services in Japan, paired with SoftBank Corp.’s operational knowhow.
SoftBank is already one of OpenAI’s largest backers and a partner in its data-centre build-out, so this is the relationship turning into a product at home. “AI is transforming cybersecurity,” OpenAI chief Sam Altman said, framing the goal as tools that “accelerate defenders.”
A launch without Altman, and without a price
The rollout was the headline, not the fine print. SoftBank put no price on the service and disclosed no contract values; everyone who attended the Tokyo presentation can apply for a free diagnosis. Altman was due to appear in person but showed up only in a short video, saying his daughter had arrived earlier than expected.
OpenAI’s chief researcher, Mark Chen, stood in.
The honest caveat is the gap between the name and the service. “Patching as a Service” sells the idea of holes sealed automatically; what is shipping is closer to an AI-assisted audit with a remediation plan attached, and humans still apply the fixes. That is not a small thing on systems where a single missed flaw can take down a power grid or an airport.
SoftBank is betting that defenders move faster with AI than without it. The bet is reasonable. The proof will be in whether the holes actually get closed.
