Anthropic’s Mythos Model Found Vulnerabilities in Classified US Government Systems, Official Says
The company said the model found flaws within hours, prompting a federal restriction and a debate over whether it weakens cyber defenses.
- On Tuesday, a U.S. official confirmed that Anthropic's Mythos AI model identified vulnerabilities in secure U.S. government systems during testing via Project Glasswing, designed to protect critical software from "severe" fallout.
- Democratic Sen. Mark Warner of Virginia revealed on June 11 that the tool breached classified systems "not in weeks but in hours," attributing the finding to Gen. Joshua Rudd, head of the National Security Agency and U.S. Cyber Command.
- The administration issued a directive restricting foreign nationals from using Fable 5 and Mythos 5 models; Anthropic disabled them to comply while questioning whether the government's security concerns were warranted.
- More than 100 cybersecurity experts from Adobe and Nvidia urged the administration to lift the directive, arguing in a letter that the models are "quite good" at finding flaws and essential for cyber defense.
- This directive followed President Donald Trump's executive order establishing a framework to vet national security risks of advanced AI systems, with participation by developers remaining voluntary.
23 Articles
23 Articles
Anthropic’s Mythos model found vulnerabilities in classified U.S. government systems, official says: AP
A U.S. official told The Associated Press on Tuesday that one of Anthropic's artificial intelligence models had identified vulnerabilities in highly sensitive and secure U.S. government computer systems during a testing exercise.
Associated Press News
Anthropic’s Mythos model found vulnerabilities in classified US government systems, official says
A U.S. official says one of Anthropic’s artificial intelligence models identified vulnerabilities in highly sensitive and secure U.S. government computer systems during a testing exercise.
AI Model Breached NSA Classified Systems in Hours
“Anthropic’s flagship Mythos AI model reportedly infiltrated nearly all of the National Security Agency’s classified systems within a few hours during an authorized red-team evaluation on June 11,” Cyber Security News reports. “This incident now seems to be the main reason for a broad U.S. government directive on export controls issued the following day.”
Coverage Details
Bias Distribution
- 43% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
