Cybersecurity leaders are calling on the US government to lift restrictions on Anthropic’s advanced security models, which are designed to identify and exploit software vulnerabilities. These AI models, classified as dual-use tools, can significantly enhance defensive cybersecurity measures but also pose a risk of being used offensively. Currently, export controls require companies to limit access to such advanced AI technologies for foreign nationals and entities, shaping the landscape of cybersecurity innovation.

Anthropic: Anthropic develops advanced AI systems with a focus on safety and reliability, including the Claude family and newer Mythos and Fable model lines. In June 2026, the company suspended global access to its latest Fable 5 and Mythos 5 models to comply with US government export controls imposed for national security reasons. Cybersecurity leaders have since urged the lifting of these curbs, arguing they limit the industry’s ability to defend against AI-powered digital threats.

Security: AI models with strong capabilities for identifying and exploiting software vulnerabilities are considered dual-use tools that support both defensive cybersecurity efforts and potential offensive applications.
Regulation: The US government has applied export controls to advanced AI models, requiring companies to restrict access by foreign nationals and entities.