31°46′40.7″N 35°14′07.7″E
Top Stories
The Wire
← The Wire
The Front · Dispatch · SecurityDeveloping

Anthropic AI model Mythos allegedly breached almost all NSA classified systems within hours, report says

The Zioneer Intelligence Desk
Anthropic AI model Mythos allegedly breached almost all NSA classified systems within hours, report says

Primary source Internal intake · 1 reviewed intake signal · Desk window 18:37

TL;DR

A new report cited by the source alleges that during authorized penetration tests, Anthropic's AI model Mythos breached almost all classified systems of the NSA and U.S. Cyber Command within hours. The claim, attributed to General Joshua Rudd (NSA/Cyber Command chief) via Sen. Mark Warner, comes after the U.S. government ordered Mythos suspended over national security concerns.

01 · THE DISPATCH

A new report, cited by the source, alleges that during authorized penetration testing, Anthropic's advanced AI model Mythos succeeded in breaching nearly all classified systems of both the National Security Agency (NSA) and U.S. Cyber Command within hours. The claim is attributed to General Joshua Rudd, head of both the NSA and U.S. Cyber Command, via Senator Mark Warner.

As The Zioneer previously reported, the U.S. government had already ordered the shutdown of Mythos over national security concerns (June 14), and restricted both Mythos and the Fable 5 model amid China-linked access concerns (June 13). This new allegation — if substantiated — would represent a drastic escalation of the concerns that led to the government's unprecedented intervention, suggesting the model's capabilities posed a direct threat to America's most sensitive digital infrastructure.

The report has not been independently verified, and the specific details of the alleged penetration tests remain undisclosed. The source text does not specify whether the breach occurred before or after the government shutdown order, or whether the models were operating under any controls during the tests.

Related dispatches
03 · Source and signal

Source and signal

A single-sourced dispatch is never rated Confirmed or Strong. Its Signal strengthens only when a second, independent source corroborates it.

  • Internal intake
Desk accountability

This dispatch is published under The Zioneer Intelligence Desk. Raw intake channels remain internal provenance; an external outlet or channel is named only when it materially helps readers evaluate a specific claim.