{"id":122629,"date":"2026-07-29T10:51:13","date_gmt":"2026-07-29T10:51:13","guid":{"rendered":"https:\/\/www.europesays.com\/ai\/122629\/"},"modified":"2026-07-29T10:51:13","modified_gmt":"2026-07-29T10:51:13","slug":"microsoft-unveils-ai-security-stack-low-cost-cyber-model-2","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/ai\/122629\/","title":{"rendered":"Microsoft Unveils AI Security Stack, Low-Cost Cyber Model"},"content":{"rendered":"

\n Artificial Intelligence & Machine Learning<\/a>
\n ,
\n Next-Generation Technologies & Secure Development<\/a>
\n ,
\n The Future of AI & Cybersecurity<\/a>\n <\/p>\n

Project Perception Combines AI Agents With New MAI-Cyber-1-Flash Model<\/p>\n

Emilia David<\/a> \u2022
\n July 27, 2026 \u00a0 \u00a0 <\/p>\n

\"Microsoft
\n Image: Shutterstock <\/p>\n

Microsoft launched Monday a new security agent stack and a cyber-capable model it says can beat Mythos but costs half as much.<\/p>\n

See Also: OnDemand | Security Operations in the Age of AI<\/a><\/p>\n

The move lets Microsoft customers tap one platform for continuous artificial intelligence-powered security surveillance and protection without having to build their own agents or contract with other AI providers for access to its top-tier cyber models.<\/p>\n

Project Perception, the agent platform, brings together a team of specialized agents and models for a continuous understanding of an organization’s security environment. It integrates<\/a> its multi-model security agent harness MDASH and a new cybersecurity model called MAI-Cyber-1-Flash.<\/p>\n

Hayete Gallot, executive vice president of Microsoft Security, said in a briefing that attackers now have an advantage against enterprises, so companies need the ability to defend themselves at the same scale and speed.<\/p>\n

“The security equation is much more complicated than it used to be,” Gallot said. “Our customers have to make a choice between innovation and safety, and that is not acceptable.”<\/p>\n

She said Project Perception represents a new kind of cybersecurity stack that eventually leads to a more automated security process.<\/p>\n

Gallot said it’s easiest to think of Perception as running three different kinds of agents: Red agents that look for attack paths, blue agents that defend and investigate, including determining what a meaningful risk is, and green agents that fix enterprise environments and harden their security postures.<\/p>\n

A user just has to ask a question on Perception – for example to assess vulnerabilities to certain attacks – and the agents begin choosing the best agents to complete the task. An agent will look into the company’s code base to find potential vulnerabilities, pass it to a second agent to build that attack path and simulate it, and then another agent writes custom vulnerability alerts while patches are fixed.<\/p>\n

“We see across identities, endpoints, applications, data, clouds and AI systems, providing broad visibility across the digital estate,” Gallot said in a blog post. “Equally important, we can help customers take action across those environments. Combined with decades of security research, threat intelligence and real-world operational experience defending organizations, these capabilities shape how Project Perception reasons, prioritizes and responds.”<\/p>\n

Perception uses a multi-model approach, meaning its agents are not confined to using specific models, whether these are from Microsoft, OpenAI or other LLMs hosted by Microsoft’s model garden. The agents will deploy the model best suited for the task.<\/p>\n

But even though Perception has its pick of the top AI models available, Microsoft still trains its own foundation models. As part of Project Perception, Microsoft released its first cyber model, MAI-Cyber-1-Flash. It’s a small, code-heavy model based on the company’s MAI-Thinking-1 LLM that trained<\/a> on the wealth of security data Microsoft amassed throughout the decades. The model integrates inside the MDASH harness and powers it along with OpenAI’s GPT-5.4.<\/p>\n

A blog post announcing MAI-Cyber-1-Flash said<\/a> the model, with GPT-5.4, outperformed other cyber-capable models such as Mythos 5, GPT-5.5 Cyber and GPT-5.6 Sol in the CyberGym evaluation. The MDASH harness scored 95.95%, compared with GPT-5.5 Cyber’s 85.6% and Mythos 5’s 83.8%.<\/p>\n

Microsoft AI CEO Mustafa Suleyman said during the briefing that its new model not only performs strongly against other models, but costs half the price too. The company did not provide pricing information for the model.<\/p>\n","protected":false},"excerpt":{"rendered":"Artificial Intelligence & Machine Learning , Next-Generation Technologies & Secure Development , The Future of AI & Cybersecurity…\n","protected":false},"author":2,"featured_media":120765,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[11],"tags":[10965,10649,420,7829,24273,24276,24275,24274,24277,24278,24279,24280,7514,7508,7511,7513,7512,7510,7509,320,7828,5990,2112,24281],"class_list":["post-122629","post","type-post","status-publish","format-standard","has-post-thumbnail","category-microsoft","tag-anti-money-laundering","tag-authentication","tag-azure","tag-azure-ai","tag-bank-information-security","tag-bank-information-security-regulations","tag-bank-regulations","tag-banking-information-security","tag-fdic","tag-fincen","tag-gao","tag-glba","tag-identity-theft","tag-information-security","tag-information-security-articles","tag-information-security-events","tag-information-security-news","tag-information-security-webinars","tag-information-security-white-papers","tag-microsoft","tag-microsoft-ai","tag-phishing","tag-risk-management","tag-sarbanes-oxley-sox"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/122629","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=122629"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/122629\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/120765"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=122629"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=122629"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=122629"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}