When China Gets Its Own Mythos (www.foreignaffairs.com)

Anthropic's recent announcement of its AI model, Claude Mythos, has raised alarms in the cybersecurity community, as it has demonstrated the capability to identify and exploit software vulnerabilities with unprecedented efficiency—outpacing skilled human experts. Notably, Mythos discovered a 27-year-old flaw in a secure operating system, highlighting a shift in the cybersecurity landscape where advanced AI can perform tasks once exclusive to elite specialists. This development underscores the urgency for the U.S. and its allies to fortify their defenses against a potential surge in AI-enabled cyberattacks, especially as adversaries like China are rapidly advancing their own AI capabilities. With models like Mythos capable of automating complex vulnerabilities exploitation, the stakes are significantly higher; an adversary could potentially execute widespread coordinated cyberattacks on critical infrastructure, such as power and water systems, more efficiently than ever before. Current U.S. cybersecurity measures may not suffice, as the window for enhancing defenses is closing. As U.S. firms restrict access to such powerful models for defensive purposes, there is mounting concern over adversaries gaining similar capabilities through illicit means or advancements in their AI technologies, posing not just a strategic threat but a pressing risk to national security.