CISA now has full Mythos Preview access, people familiar say
CISA now has full Mythos Preview access, people familiar say
Publish Date: 2026-06-17 17:21:00
Source Domain: www.nextgov.com
Summary:
The Cybersecurity and Infrastructure Security Agency (CISA) has received access to Anthropic’s flagship Mythos Preview model, an AI tool that can identify vulnerabilities in computer networks. However, the White House’s Office of the National Cyber Director has not specified how CISA should use this model, reflecting a broader issue where federal tech leaders have felt inadequately briefed on its implementation and use for vulnerability scanning. This move comes amidst an intense debate in the cybersecurity community about how advanced AI models like Mythos could reshape both defensive and offensive cybersecurity operations. While the Mythos Preview model has been distributed to select organizations through Project Glasswing, efforts have been made to control it due to its potential to either significantly boost defenders’ capabilities or adversaries’ offensive hacking operations. Meanwhile, the successor model, Mythos 5, has been banned through export controls.
Key Points:
- CISA now has access to Anthropic’s Mythos Preview model but lacks clear usage guidelines.
- Federal agencies have reported inadequate briefings from the Office of the National Cyber Director on using AI for vulnerability scanning.
- The potential of AI models like Mythos to both enhance and threaten cybersecurity operations is a growing concern.
- Anthropic’s advanced AI model, Mythos 5, was recently banned due to export control mechanisms.
- The shift in the management and use of advanced AI models reflects changing priorities in cybersecurity practices under the current administration.
