1 Day
Source:
وكالة الأنباء القطرية
Anthropic boosts cybersecurity AI access
Anthropic announced the expansion of cybersecurity teams' authority to access advanced AI models, including the "Claude Mythos" model, with reduced restrictions. The company integrated the "Project Glasswing" and "Cybersecurity Verification Program" into a three-tier system that enables the use of these models in defensive tasks, penetration testing, and critical systems such as power grids and aviation. Tests showed that the "Mythos" models achieved varying degrees of success, with the "Opus 5.5" model managing to surpass four attempts within the defense level and 34 within the penetration team's level, reflecting high capabilities in cybersecurity while highlighting the need for risk management to prevent misuse.
Size: 120%Text Size
Notice: This Is an AI-Generated Summary
Comments (0)