Anxious Security Pros Watch as Anthropic, OpenAI Expand Access to Frontier AI Models

### Executive summary This article analyzes the cybersecurity implications of frontier AI models—particularly Anthropic’s Claude Mythos and OpenAI’s GPT-5.5-Cyber—highlighting their ability to rapidly find vulnerabilities and produce exploits, the controlled access program (Project Glasswing) to limit misuse, concerns about overwhelming organizations with machine-generated vulnerability disclosures, and recent U.S. executive-order guidance seeking pre-release government review of high-risk models; it is a policy and risk discussion rather than a report of an active incident.