US cybersecurity agency using Anthropic’s Mythos to audit govt software: Report
Despite the Trump administration's feud with Anthropic, the US cybersecurity agency CISA is reportedly using Mythos to audit government software for bugs that could leave the door open for spies and cybercriminals.

The US cybersecurity agency is using Anthropic's AI model Mythos to audit government software, according to Reuters.
The Cybersecurity and Infrastructure Security Agency (Cisa) is using Mythos to scan code repositories in government systems for bugs that could leave the door open to spies and cybercriminals.
Cisa's Attack Surface Evaluation team is leading the scan and has already found "a large number of vulnerabilities. The team is dedicated to digital security assessments and hacking exercises across the government. The team is dedicated to digital security assessments and hacking exercises across the government.
Mythos is a highly capable AI model that can identify unknown flaws in computer networks and systems. While this could allow cybersecurity professionals to find bugs and vulnerabilities and fix them, it could also enable cybercriminals to mount catastrophic attacks. That is why Anthropic delayed the public release of the model and began with a limited rollout to technology companies and banks so they could prepare their digital infrastructure.
The report highlights that the US government continues to use Anthropic's models despite the Donald Trump administration's ongoing feud with the company.
In February, the administration barred the company from the defence supply chain after it refused to grant the Department of Defense unrestricted access to its models for “any lawful use” without strict safety safeguards against domestic surveillance and fully autonomous targeting. The ban was blocked by a court in March.
Previously, Axios reported that the US cyber-spy agency, the National Security Agency (NSA), had been using Mythos since April despite the administration blacklisting it in February. The report said that NSA analysts had tested Mythos in classified settings and were impressed by its capabilities.
When Anthropic released the public version of Mythos, dubbed Fable, the Trump administration told the company to bar non-Americans from accessing it. The ban was lifted last week.

Pakistan's deep state weaponising LeT to counter Asim Munir critic Maulana Fazlur Rehman: Report
No premium queues, 30-minute target: What changes for Indian passport applicants in UAE
'We don't need straits': Trump says Iran war isn't about Hormuz as US strikes for 12th night
Iran claims to stop three oil tankers transiting Strait of Hormuz
Globe at a glance: From Houthis striking Saudi tankers in Red Sea to wildfires in Europe
