Anthropic's Mythos model found vulnerabilities in classified US systems, official says

SecurityWeek

Anthropic's Mythos AI identified vulnerabilities in highly sensitive US government computer systems within hours during a testing exercise with intelligence agencies, a US official told AP. The testing was part of Anthropic's Project Glasswing initiative. Sen. Mark Warner had previously mentioned that Mythos 'broke into almost all of our classified systems' in hours.

AnthropicMythos

See 2 more sources

Reuters has now added more context to last week’s Mythos reporting. According to AP, Anthropic’s...7 days agoKimmonismus

Mythos was not trained on 'hacking'. Other Ai labs also will reach Mythos-level capabilities in the future9 days agoHyperspaceAndBeyondDiscuss

SecurityWeek

Jun 24, 3:29 AM