Home / Technology / AI Models Show Major Leap in Vulnerability Discovery
AI Models Show Major Leap in Vulnerability Discovery
14 May
Summary
- AI models like Claude and GPT-5.5 excel in cybersecurity testing.
- Frontier models demonstrate significant progress in finding vulnerabilities.
- Microsoft's AI agent discovered 16 CVEs in recent updates.
AI models are showing remarkable progress in enhancing cybersecurity. On Wednesday, the AISI reported that Anthropic's Claude Mythos Preview and OpenAI's GPT-5.5 models performed significantly better than expected in cybersecurity testing.
Independently, XBOW released data indicating that "frontier models" have advanced substantially in their ability to discover vulnerabilities. This suggests a new era of AI-powered security.
Microsoft also announced that its multi-model agentic setup, MDASH, successfully identified 16 CVEs during this week's Patch Tuesday updates. MDASH also leads in the CyberGym security evaluation framework.
Disclaimer: This story has been auto-aggregated and auto-summarised by a computer program. This story has not been edited or created by the Feedzop team.
