software vulnerabilities

Winsage
July 18, 2026
Microsoft is developing a security tool called Project Perception, aimed at helping organizations identify and fix software vulnerabilities using artificial intelligence. Set to launch this month, it will utilize AI models from Microsoft, OpenAI, and Anthropic to offer extensive coverage while being cost-effective. The tool features a model router to select the best AI model for specific tasks, optimizing performance and reducing operational costs. Led by Hayete Gallot, Microsoft is shifting towards AI-focused products, recognizing the potential for substantial returns from enterprise investments in AI. Project Perception aims to compete in the AI-driven vulnerability detection market, although its availability may be limited due to concerns about misuse. Microsoft recently addressed around 570 vulnerabilities using an AI tool called MDASH, suggesting a growing role for AI in cybersecurity.
Winsage
July 12, 2026
Microsoft is integrating artificial intelligence into its vulnerability detection processes for the Windows operating system to enhance security. This will lead to more frequent security updates during monthly Patch Tuesday releases. The company aims to address the rise in AI-driven exploits and is refining its secure software development model to combat evolving tactics used in AI-driven attacks. While AI will assist in identifying vulnerabilities, human oversight will remain essential, with developers reviewing code and validating AI-generated findings before deploying updates.
Winsage
June 3, 2026
Microsoft's Build developer conference featured the introduction of Microsoft Scout, an AI tool designed to automate tasks by integrating with Microsoft 365 data. The conference also announced "Codename MDASH," a multi-model agentic scanning system for identifying and fixing software vulnerabilities. Additionally, the Surface RTX Spark Dev Box was unveiled, a compact developer PC with Nvidia's RTX Spark chip, up to 128GB of memory, preloaded with Windows 11 Pro, and tailored for developers. The Windows Dev Kit 2023 was also highlighted, powered by Qualcomm’s Snapdragon 8cx Gen 3, facilitating the development of Arm-based Surface devices with improved code translation technology.
Winsage
May 20, 2026
Bitdefender's research highlights the use of Microsoft's MSHTA utility in malware attacks, noting its default activation in Windows systems. Cybercriminals exploit MSHTA to execute malicious scripts under the guise of legitimate processes, linking it to various malware families like LummaStealer and PurpleFox. The study reports a rise in MSHTA-related detections, indicating a shift towards "living-off-the-land" tactics that utilize legitimate tools to evade security alerts. Social engineering is identified as a common entry point for attacks, employing deceptive methods such as fake software downloads and phishing links. MSHTA can retrieve and execute additional payloads through multi-stage chains, complicating detection efforts. The attacks target sensitive information, including credentials and financial data, and the continued presence of MSHTA poses risks as it allows threat actors to conceal malicious actions. To mitigate these threats, organizations are advised to restrict or disable legacy scripting tools and exercise caution with untrusted downloads. The report emphasizes the challenge of detecting unusual behaviors associated with legitimate utilities in the context of cyber threats.
Winsage
May 15, 2026
Microsoft has released an extensive update for Azure Linux 3.0 and CBL Mariner 2.0, addressing 191 open-source Common Vulnerabilities and Exposures (CVEs) across various technologies, including the Linux kernel, Go runtime, Apache httpd, PHP, CoreDNS, Valkey, Ruby, GnuTLS, Apache Thrift, Node.js, Rust, Java implementations, Vim, Postfix, Expat, Nmap, Prometheus, KEDA, and PgBouncer. Additionally, Microsoft has fixed a critical vulnerability (CVE-2026-41103) in its Single Sign-On (SSO) Plugin for Jira and Confluence, which allows an attacker to forge a Microsoft Entra ID identity through a manipulated SAML response; however, patching this vulnerability is the responsibility of the users of Atlassian's platforms.
Winsage
May 14, 2026
Microsoft has introduced MDASH (Multi-Model Agentic Scanning Harness), a security solution that uses over 100 specialized AI agents to identify software vulnerabilities. On May 12, 2026, MDASH identified 16 new vulnerabilities (CVEs) in the Windows networking and authentication stack, four of which were critical, including remote code execution vulnerabilities in tcpip.sys, ikeext.dll, netlogon.dll, and dnsapi.dll. Ten of these vulnerabilities can be accessed over the network without authentication. MDASH operates through a four-stage pipeline: analyzing source code, scrutinizing for suspicious elements, debating the exploitability of issues, and attempting to exploit vulnerabilities. The system is model-agnostic and allows integration of new models and domain-specific knowledge. MDASH scored 88.45 percent on the CyberGym benchmark, ranking first among competitors, although the comparison may not be entirely fair as it contrasts a comprehensive framework with individual models. The models used to achieve this score are not specified. MDASH is supported by Microsoft's Autonomous Code Security Team and is currently in a limited private preview for select customers.
Search