🆘 Governments Worldwide Mobilize to Counter Cyber Threat from Anthropic's AI System Mythos

The U.S. House Homeland Security Committee held a closed briefing with representatives from Anthropic, while Palo Alto Networks warned that such attacks could become the "new normal" within months.

During internal testing, engineers "without specialized security training" tasked Mythos with finding exploitable vulnerabilities in software overnight—and by morning, the model had prepared full, ready-to-use exploits.
For the first time, both AISI cyber simulation tasks were completed, including one that no previous model had been able to solve.

What used to take months, AI does in minutes) 🔥

⚡️ Acute security issue here

🫥 UNSERO: Digital Horizon