tech
Anthropic says its latest AI model can expose weaknesses in software security
AI company says purpose of its Claude Mythos model is to bolster defenses against hacking in common applications
TL;DR
- Anthropic's new AI model, Claude Mythos, can find thousands of previously undiscovered software vulnerabilities.
- The model is not being released to the public due to concerns about hackers exploiting its capabilities.
- Anthropic is partnering with cybersecurity firms and tech giants in a project called 'Glasswing' to use Mythos for defense.
- The project aims to bolster defenses against hacking and leverage AI to find and fix vulnerabilities at an unprecedented pace.
- The AI's ability to identify vulnerabilities is surpassing human capabilities, leading to urgent calls for enhanced cybersecurity measures.