Intel Corporation is announcing participation in Project Glasswing, launched by Anthropic, that deploys the new Claude Mythos Preview AI model to proactively identify and remediate critical vulnerabilities across legacy technology. Intel joins a collective of industry leaders partnering to put the defensive potential of advanced AI to work before adversaries can exploit the same capabilities.
Intel plays a unique role in this effort, working at the forefront of silicon and software. From designing and manufacturing the silicon that powers the world's computing infrastructure to the software and platform security features embedded in our processors, Intel products represent a foundational layer of trust that the entire industry depends upon.
AI models are advancing to the point where they are poised to discover security vulnerabilities at a pace and scale not seen previously. The same capabilities that offer the potential to make this model powerful in the hands of defenders will inevitably reach adversaries.
Intel has long approached security as a mindset, not a milestone. We design with security in mind: protections built into silicon, establishing hardware as the root of trust, and products developed, manufactured and maintained according to industry-leading security assurance practices. Project Glasswing applies that philosophy to the AI era: getting ahead of emerging threats by harnessing AI's offensive capabilities exclusively for defense, at scale.
“Security starts with Intel. From protections rooted in silicon through every layer of the stack, Intel plays a foundational role in securing the world’s computing infrastructure. The emergence of AI models capable of discovering vulnerabilities at machine speed and scale is not a future risk — it is already a reality. Addressing this requires coordinated action across the industry and Project Glasswing represents that collective effort. Intel is proud to contribute, recognising that strengthening the security of global computing infrastructure is a shared responsibility — and one that no single organisation can address alone.”
While this engagement is new, our focus on AI confidentiality and security has deep roots, including years of focus on securing AI workloads through software such as Intel® Trust Domain Extensions and Intel® Software Guard Extensions, offensive security research from the INT31 team into cutting-edge validation techniques, and enhancing platform security solutions, many of which are detailed in the 2026 Intel® Platform Security Report.
This same emphasis led to Intel joining as a founding member of the Coalition for Secure AI (CoSAI), working across the industry to share best practices for secure deployment and collaborate on AI security research and product development.
Intel will use access to Claude Mythos Preview to scan drivers, system firmware, and embedded firmware critical to our ecosystem, as well as exploratory work on select hardware. Vulnerabilities discovered through this work will be responsibly disclosed and remediated in coordination with Anthropic's established reporting processes and Intel's existing Product Security Incident Response Team (PSIRT).
Intel anticipates that guidance on securing the software development lifecycle, supply-chain integrity, and firmware security will be among the most consequential outputs of this collaboration for the broader industry.
Share Your Feedback
We want to hear from you. Send comments, questions, and feedback to the INT31 team.
About the Author
Dhinesh Manoharan is vice president and head of INT31 Security Research. In this role, he drives ‘Security for AI’ and ‘AI for Security’ strategy across the company. He has the privilege of leading an eclectic global team of systems-level security researchers focused on proactive research of threats, zero day vulnerabilities, mitigations and exploits in the Client, Edge, Datacenter and AI ecosystems to keep billions of people across the world secure in their digital lives. He works very closely with world-wide customers, industry consortia, academia collaborators and government agencies on contemporary cybersecurity initiatives. Dhinesh holds an MS degree in Computer Science from Portland State.