Anthropic Reveals Mythos Access to NSA, Excludes CISA

At the recent Cloud Next 2026 event, Anthropic announced that its advanced vulnerability detection model, Claude Mythos, will be available to the National Security Agency (NSA) while explicitly excluding the Cybersecurity and Infrastructure Security Agency (CISA) from access. This decision highlights the intricate dynamics between different U.S. governmental bodies regarding cybersecurity resources and capabilities.

The implications of this decision are significant, as granting exclusive access to the NSA could enhance national security efforts while simultaneously raising concerns about the cybersecurity capabilities of CISA. By limiting access to its technology, Anthropic underscores a potential rift in U.S. cybersecurity policy, which may lead to a dependency on specific agencies for critical security technologies, thus impacting the broader landscape of national AI strategies.