Anthropic opens Mythos and other AI models to vetted security teams
Anthropic is expanding its Cyber Verification Program, giving approved security teams access to its most capable models with safeguards tailored to their work. Reuters reports the move follows Project Glasswing partners finding at least 129,000 verified software vulnerabilities between April and July, including more than 33,000 rated critical or high severity.
Bottom line — Anthropic’s three-tier programme gives defenders broader access while reserving the fewest restrictions for authorised tests of safety-critical systems.
Go deeper 6
-
Reuters reports Anthropic’s own open-source scans found 5,500 further vulnerabilities between April and October.
-
Anthropic says the reported totals are likely an undercount and estimates the programme’s true impact may be at least five times higher.
-
According to Reuters, the Defense tier covers work such as incident response and malware analysis; the Red Team tier adds authorised penetration testing.
-
National Technology reports that the Specialized tier is for organisations authorised to test safety-critical systems, including power grids, flight systems and telecoms networks.
-
National Technology says Anthropic reviews organisations in the Specialized tier with the US government; Reuters reports existing Glasswing members will transition into it.
-
National Technology reports that safeguards remain against actions such as ransomware deployment and activity that could cause physical harm or mass disruption.