Microsoft announced the launch of two AI-powered security tools designed to help organizations identify and mitigate cybersecurity vulnerabilities. The announcement occurred on Monday, less than a week after OpenAI’s security models were compromised during a breach at startup Hugging Face involving thousands of automated actions that exploited a zero-day flaw.
The first tool is MAI-Cyber-1-Flash, a specialized AI model trained to identify and remediate software security weaknesses. Built on Microsoft’s MAI-Thinking-1 platform, the model leverages Microsoft’s extensive experience in vulnerability patching and security incident response across its customer base. Microsoft stated it processes over 1 trillion security signals daily and maintains insights from 1.6 million customers. The company integrated MAI-Cyber-1-Flash into MDASH, a multi-model scanning system that combines 100 security-trained AI agents to discover exploitable bugs in applications. According to Microsoft, the updated MDASH achieved a 96 percent score on the CyberGYM benchmark test, surpassing Anthropic’s Mythos, Google Gemini, and OpenAI’s GPT offerings by significant margins. The company also reduced the operational cost of MDASH to half its previous pricing.
Microsoft’s second offering, Project Perception, comprises specialized AI agents that perform security functions including vulnerability discovery, risk assessment, and corrective action implementation. The platform automatically selects appropriate AI models based on task requirements, effectiveness metrics, and cost considerations. Microsoft stated that Project Perception can handle 90 percent of typical security tasks at lower costs than competing solutions.
Microsoft characterized the tools as responses to evolving threats in cybersecurity, noting that organizations increasingly struggle to manage complex security operations amid accelerating attack speeds and scales. The company emphasized that security teams face challenges in synthesizing security signals and risk assessments across large datasets. Both tools currently remain in preview mode pending broader availability.
Article Attribution | Read More at Article Source
Article summary produced by Claude AI