AI Is Speeding Up Both Attacks and Software Development: Why Context Matters More Than Alerts

Wiz Research says frontier AI models and AI-assisted development are placing new pressure on application security. On one side, AI models can autonomously discover vulnerabilities and generate working exploits, shrinking the window from discovery to exploitation from days to minutes. On the other, teams are shipping code faster than ever, which amplifies existing risks and widens the attack surface.

According to the article, this double acceleration breaks traditional workflows, where people manually reviewed scanner alerts, tracked down code owners and assigned tickets. Wiz argues that Application Security Posture Management (ASPM) is therefore more important than before, but it must do more than collect alerts in one place. It needs to use context from across the application lifecycle to help teams focus on risks that can actually be exploited.

The article uses React2Shell (CVE-2025-55182), a critical remote code execution flaw affecting Next.js applications, as an example. For an organisation heavily invested in Next.js, a flaw in a core component can generate thousands of separate findings across hundreds of code repositories and container images. Without context, security teams face a huge backlog with no clear way to tell which instances are truly exploitable. This is the first part of a series; later parts will cover prioritisation and remediation.