ai
National Security Agency Using Anthropic Mythos AI Model for Hacking
The National Security Agency is using Anthropic's unreleased Mythos AI model for offensive cyber operations. The model was developed to detect and exploit software vulnerabilities and is being provided to select institutions for defensive purposes.
Jun 5, 9:26 AM·1m read3 sources
SemaforAudio version
Tap play to generate a narrated version.
Mythos has not been publicly released due to its capabilities in detecting and exploiting software vulnerabilities. The model is instead being offered to select institutions to help them build defenses against such threats. The agency is reportedly applying the model for offensive purposes while the U.S. government maintains a separate legal action against Anthropic.
AI agents now account for most web traffic, according to Cloudflare data cited in the report. The figure arrived a year ahead of earlier projections. A portion of this traffic consists of malicious bots. Iran has been identified as a location with elevated activity of this type. Open-source AI models trail the most advanced systems by less than one year in capability.
Transparency
How sources framed this
These outlets didn't split into competing frames — coverage was uniform.
CorroborationLimited · 3 sources
