Anthropic's Project Glasswing, which uses AI to find software vulnerabilities, is facing scrutiny from cybersecurity firm VulnCheck. While Claude has identified nearly 38,000 potential bugs over seven months, only 217 have been publicly recorded as fixed. This discrepancy has led to questions about the program's effectiveness and the ability of open-source maintainers to keep up with the sheer volume of AI-generated bug reports. Anthropic acknowledges that manual review is a bottleneck and has introduced a service for direct code scanning by Claude to expedite the process. AI
IMPACT Highlights challenges in AI-driven vulnerability discovery, specifically the bottleneck in human review and the strain on open-source maintainers.
RANK_REASON The article discusses the effectiveness and impact of a specific AI initiative (Project Glasswing) based on a third-party analysis, rather than announcing a new release or product.
- Anthropic
- Apple Inc.
- AWS
- Claude
- curl
- Daniel Stenberg
- GitHub
- Microsoft
- Nvidia
- OSS Scanner
- Project Glasswing
- VulnCheck
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
