Mythos: A New Era in Cyber Defense Amidst Government Struggles

The U.S. cyber defense agency CISA uses Anthropic's AI model Mythos to audit government software, revealing vulnerabilities. This move marks the government's interest in AI tools amidst Anthropic's ongoing issues with U.S. authorities regarding AI's use in weaponry and surveillance, prompting complex relations with the White House.

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has begun employing Anthropic's AI model, Mythos, to audit government software. This initiative, aimed at identifying vulnerabilities that could attract foreign spies and cybercriminals, reflects growing governmental interest in AI implementations despite Anthropic's rocky relationship with the White House.

According to sources, CISA's Attack Surface Evaluation team, responsible for cybersecurity assessments across government entities, has already detected numerous vulnerabilities during their scans. However, specifics remain undisclosed, and neither Anthropic nor CISA have provided further details about the scope of the audits or the nature of the vulnerabilities uncovered.

Anthropic's relationship with the U.S. government has been fraught, hitting a low point in February when the company stood firm on AI safeguards that prevented misuse. This stance led the Pentagon to label it as a supply-chain risk. Tensions have since eased, especially after the National Security Agency began utilizing Mythos. Nevertheless, the model's use remains a point of contention, with security measures causing temporary global shutdowns.