Anthropic’s Project Glasswing uncovers over 10,000 software vulnerabilities using Claude AI
Anthropic's Project Glasswing has identified over 10,000 software vulnerabilities using its Claude Mythos Preview AI model. The vulnerabilities include decades-old zero-day flaws in major operating systems and browsers. This initiative highlights the growing gap between the speed of AI in discovering vulnerabilities and the slower pace of human patching efforts.
- ▪Project Glasswing was launched on April 7, 2026, and within six weeks, it uncovered a significant number of security flaws.
- ▪The vulnerabilities discovered include a 27-year-old flaw in OpenBSD and a 16-year-old flaw in FFmpeg.
- ▪Claude Mythos Preview demonstrated the ability to chain vulnerabilities and develop working exploits autonomously.
- ▪Access to Claude Mythos Preview is restricted to vetted partners like AWS, Apple, Google, and Microsoft.
- ▪The findings reveal a bottleneck in the cybersecurity industry, where AI can find vulnerabilities faster than humans can patch them.
Opening excerpt (first ~120 words) tap to expand
Anthropic’s Project Glasswing uncovers over 10,000 software vulnerabilities using Claude AI The AI lab's unreleased Claude Mythos Preview model found decades-old zero-day flaws in major operating systems and browsers in just one month. Share Add us on Google by Editorial Team May. 23, 2026 window.sevioads = window.sevioads || []; var sevioads_preferences = []; sevioads_preferences[0] = {}; sevioads_preferences[0].zone = "01f21ccf-2092-46b1-9ac7-8c44cc782e0f"; sevioads_preferences[0].adType = "native"; sevioads_preferences[0].inventoryId = "c5700508-581b-472c-8fdd-a931cdbfc8e1"; sevioads_preferences[0].accountId = "1e47efc1-ec2d-4fca-a8b9-354e249e5095"; sevioads.push(sevioads_preferences); Anthropic just demonstrated what happens when you point a sufficiently powerful AI model at the…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Crypto Briefing.
Discussion
0 commentsMore from Crypto Briefing
