AI Security Advances Underscore Persistent Need for Runtime Application Testing

According to a recent LinkedIn post from StackHawk, recent market reactions to Anthropic’s AI security tools have highlighted investor uncertainty around the impact of autonomous code analysis on cybersecurity vendors. The post notes that cybersecurity stocks initially fell after the launch of Claude Code Security in February, but later rallied following Anthropic’s Project Glasswing announcement with major cloud and security partners.

The LinkedIn post highlights that Project Glasswing involves a $100 million cloud credit coalition including Microsoft, AWS, Apple, Google, and CrowdStrike, and experimentation with Claude Mythos Preview for automated vulnerability discovery. According to the commentary, these developments underscore rapid advances in AI-driven code analysis, while also raising questions about how similar capabilities may be leveraged by both defenders and attackers.

StackHawk’s post argues that despite growing sophistication in AI code analysis, these tools remain fundamentally different from runtime security testing. The author suggests that AI models reviewing source code may miss authorization flaws such as broken object-level authorization (BOLA) and broken function-level authorization (BFLA), which manifest only in live application behavior rather than in static code.

For investors, the analysis implies that AI code analysis could pressure traditional static analysis vendors while creating opportunities for companies focused on runtime and application-level testing. The post further suggests a potential industry gap where even advanced AI systems may not fully address authorization vulnerabilities, which could sustain demand for specialized runtime security platforms like those in StackHawk’s segment.