Microsoft on Monday launched its first cybersecurity-specialized mannequin alongside a brand new AI cybersecurity platform at a small occasion in San Francisco, taking a giant swipe at main gamers within the area — particularly Anthropic, Google and OpenAI.
The corporate describes MAI-Cyber-1-Flash as a mannequin that’s constructed “to search out difficult vulnerabilities in advanced codebases.” The mannequin is constructed to animate MDASH, Microsoft’s harness devoted to software program vulnerability identification and remediation.
The brand new safety platform is dubbed Notion, and it’s designed to deploy teams of agents to help with and automate varied safety workflows, together with figuring out and remediating bugs. The platform can even combine with MDASH.
The corporate claims MAI-Cyber-1-Flash is considerably extra highly effective (and more cost effective) than competitor fashions, based mostly on its efficiency on a longtime AI cybersecurity benchmark.
“We’re very very excited to announce our outcomes,” stated Mustafa Suleyman, the co-founder of DeepMind and present CEO of Microsoft AI. “We now have MAI-1 Cyber Flash binded [sic] with GPT 5.4 inside the MDASH harness — which beats out Gemini, GPT 5.5 Cyber, GPT 5.6 Sol, and Mythos 5 on Cyber Health club, which is the first benchmark that all of us use. The golden benchmark.”
“We’re delivery this into manufacturing instantly,” he added.
Noting that hackers are more and more utilizing AI of their cyberattacks, Hayete Gallot, Microsoft’s vp for safety, described Notion as a method for enterprise defenders to “defend towards AI with AI on the scale and velocity that the attackers have.”
Notion makes use of agentic purple groups, blue groups, and inexperienced groups. The purple groups can present detailed simulations of potential assaults — offering context about potential risk actors and the probably vulnerabilities that they could exploit. Blue groups are devoted to detecting and triaging present bugs, whereas inexperienced groups take “corrective actions” towards these bugs.
Dave Weston, the lead engineer for Notion, described the platform as a large effectivity improve for company defenders. “We’ve gone from this taking hours and hours of guide work from a number of specialised people throughout the safety group — appsec hunters, remediation engineers, you title it — and in minutes, we’ve a repair for all of this. Not solely can we uncover the problems and prioritize them, however we’ve detection, posture fixing, and even a code repair.”
Although AI has supplied new defensive capabilities to corporations, its availability to cybercriminals has given rise to a stunning array of potential threats.
Microsoft’s new safety instruments, which the corporate stated will likely be out there in preview on November 3, will enter an more and more crowded discipline of AI cybersecurity options. Earlier this yr, Anthropic launched Mythos, a safety platform that was launched to a small coterie of companion organizations by a program known as Glasswing. OpenAI has also launched its personal safety answer in Could by a program known as Day Break.
Whenever you buy by hyperlinks in our articles, we may earn a small commission. This doesn’t have an effect on our editorial independence.
