The US Cybersecurity and Infrastructure Security Agency (CISA) is deploying Anthropic's artificial intelligence model Mythos to scan government software for security flaws, three people familiar with the matter told Reuters.
The development comes in a fresh sign of federal appetite for the AI startup's tools even as Anthropic remains locked in a standoff with the White House.
CISA's Attack Surface Evaluation team, a unit that runs digital security assessments and hacking exercises across government, is using the AI to comb through government code repositories for vulnerabilities that could be exploited by foreign spies or cybercriminals, the sources reportedly said.
Two of the sources said the audits had already turned up a significant number of vulnerabilities, though they did not elaborate further, and the agency said it could not independently verify the extent of the code reviewed or the severity of the bugs identified.
Anthropic did not respond to questions about the initiative, while a CISA representative said last month it would look into the matter but did not follow up.
The report comes against the backdrop of a rocky relationship between Anthropic and Washington.
ALSO READ: Samsung's 19-Fold Suge In Profit Fails To Impress Investors After AI Chip Rally
Ties soured sharply in February after the company declined to strip away safeguards blocking its AI from being used in autonomous weapons systems or domestic surveillance, prompting the Pentagon to impose a supply-chain risk designation on the firm, a label typically reserved for foreign companies suspected of espionage links.
A judge subsequently blocked the designation, and tensions have eased since Anthropic privately released Mythos, a model regarded as highly capable at identifying and exploiting cybersecurity flaws.
National Security Agency has been using Mythos since as early as April despite the blacklist, while The New York Times reported last month that NSA analysts testing the model in classified settings came away impressed with its capabilities.
However, when Anthropic launched a public version called Fable with built-in safeguards, the White House demanded a ban on foreign nationals operating it, triggering a global suspension of the model that was lifted only last week.
The NSA and the White House did not immediately respond to Reuters' requests for a comment.
ALSO READ: US Judge Rejects Elon Musk's Attempt To Overturn Twitter Fraud Verdict
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.
