Epic Systems Joins Anthropic’s Project Glasswing to Secure Healthcare Codebase
The EHR giant is leveraging Claude Mythos for cybersecurity while deepening its clinical AI integration with UpToDate.
Epic Systems is partnering with Anthropic to enhance the security of its healthcare software through Project Glasswing, a specialized cybersecurity initiative. The collaboration was confirmed during Epic's annual Users Group Meeting (UGM), signaling a strategic shift toward using advanced large language models (LLMs) for infrastructure protection.
As part of Project Glasswing, Epic is utilizing Anthropic’s Claude Mythos model to proactively identify and remediate software vulnerabilities within its massive codebase. By applying the model's reasoning capabilities to security audits, Epic aims to harden its systems against potential exploits in an era of increasing cyber threats to healthcare infrastructure.
Expanding Clinical AI Capabilities
Beyond cybersecurity, Epic is expanding its clinician-facing AI ecosystem through a deepened partnership with Wolters Kluwer’s UpToDate. The company is integrating the evidence-based clinical reference database directly into "Art," Epic's native AI assistant for clinicians. This integration allows providers to access vetted medical guidance without leaving their primary workflow.
Furthermore, the UpToDate integration extends into Epic’s "Agent Factory." This framework enables individual health systems to develop customized AI agents that possess direct access to UpToDate’s clinical guidance, allowing hospitals to tailor AI-driven decision support to their specific patient populations and institutional protocols.
Industry Implications
These developments represent a dual-track strategy to integrate LLMs into the most widely used electronic health record (EHR) system in the U.S. While the UpToDate integration focuses on the "last mile" of clinical decision support, Project Glasswing addresses the systemic risk of software vulnerabilities. For the healthcare industry, this suggests a move toward "AI-native" security and clinical workflows where the LLM acts as both a shield for the codebase and a bridge to medical knowledge.
Future Outlook
As Epic continues to roll out these tools via the Agent Factory, the industry will be watching how health systems implement these customized agents in live clinical settings. While the technical framework for Project Glasswing is now established, the long-term impact on Epic's vulnerability remediation timelines remains to be seen. This shift toward automated security audits could potentially reduce the window of exposure for critical healthcare data, provided the Claude Mythos model maintains high precision in its vulnerability detection.