Latest / GNN Daily Breaking News / U.S. Cyber Agency Uses Anthropic’s Mythos to Audit Government Code

Transcript

U.S. Cyber Agency Uses Anthropic’s Mythos to Audit Government Code · GNN Daily Breaking News

• 0:00
  Speaker 1: The Cybersecurity and Infrastructure Security Agency SISA has begun using
• 0:05
  anthropics Advanced AI model MYTHOS to audit government code repositories,
• 0:10
  marking a significant shift in how federal systems are secured.
• 0:14
  The initiative is part of sis's Attack Surface Evaluation program,
• 0:19
  which conducts digital security assessments and simulated hacking exercises to
• 0:24
  identify vulnerabilities before adversaries can exploit them. MYTHOS is designed
• 0:29
  to scan massive codebases at machine speed, uncovering bugs and
• 0:34
  weaknesses that would traditionally take weeks for human analysts to detect.
• 0:38
  Officials say the AI has already flagged numerous flaws, though
• 0:43
  details on their severity remain classified. The model's ability to
• 0:47
  compress timelines for vulnerability detection is seen as a critical
• 0:51
  advantage in defending against foreign spies and cyber criminals. The
• 0:56
  deployment comes under Project glass Wing, a broader initiative that
• 1:00
  has expanded to include partners in more than fifteen countries.
• 1:04
  While this international collaboration strengthens collective defenses, it also raises
• 1:09
  concerns about shared exposure if adversaries gain insight into Mythos' methods.
• 1:15
  The National Security Agency NSA has also tested Mythos in
• 1:20
  classified environments, reporting strong results. Yet Anthropiic's relationship with the
• 1:25
  government has been complicated. Earlier this year, the Pentagon labeled
• 1:30
  the company a supply chain risk after Anthropic refused to
• 1:33
  remove safeguards preventing its AI from being used for autonomous
• 1:37
  weapons or domestic surveillance. A federal judge later blocked the blacklisting,
• 1:43
  clearing the path for Mythos' adoption. Beyond security, Anthropic is
• 1:48
  positioning itself for growth, having confidentially filed for a US
• 1:53
  initial public offering. The company's dual role as a frontier
• 1:57
  AI innovator and a sensitive government contrast underscores the tension
• 2:02
  between innovation regulation and national security. For CISA, Mythos represents
• 2:08
  a milestone in AI driven cybersecurity. For policymakers, it raises
• 2:14
  urgent questions about oversight, privacy, and the dual use risks
• 2:18
  of deploying powerful AI in sensitive environments. And for the public,
• 2:23
  it signals that artificial intelligence is no longer just a
• 2:27
  tool of convenience. It is becoming a frontline defense in
• 2:30
  the digital age.