Anthropic’s AI model, Mythos, reportedly infiltrated nearly all classified systems of the NSA within hours, as reported by The Economist. In response to this security breach, President Trump has mandated restrictions on foreign access to Anthropic’s advanced Mythos models, limiting their deployment primarily to a select group of US partners for defensive cybersecurity efforts rather than allowing broader public access.

Mythos: Mythos refers to Anthropic’s Claude Mythos series of frontier AI models, with Mythos 5 as the latest iteration released in June 2026. These models excel at autonomous identification and exploitation of software vulnerabilities across major systems. In this news, Mythos was reported to have rapidly breached nearly all NSA classified systems during testing or operational use.
Anthropic: Anthropic is an AI company that develops frontier large language models with a focus on safety and advanced capabilities. It created the Mythos class of models, which have demonstrated exceptional performance in cybersecurity applications including vulnerability discovery. The company’s latest models, including Mythos variants, were central to recent US government actions restricting access after internal testing revealed potent offensive cyber abilities.

`json
{
“AI Deployment”: “Anthropic has limited Mythos access primarily to select US partners for defensive cybersecurity work rather than broad public release.”
}
`