AI-Written Software Creates Hassles for Wary Security Teams

Why It Matters

The surge in AI‑written code creates new attack vectors, forcing firms to strengthen governance and invest in controls, which could impact compliance and overall risk exposure across the industry.

Key Takeaways

• 38% of security pros can keep up with AI code
• 60% say reviewing AI‑generated code is getting harder
• 78% fear AI tools could expose corporate secrets
• 73% cite supply‑chain risks from AI‑created dependencies
• Teams demand audit trails and access limits before AI adoption

Pulse Analysis

The rapid adoption of AI‑assisted coding is reshaping software development pipelines, enabling engineers to ship features at unprecedented speed. While productivity gains are evident, the volume of automatically generated code is outpacing the capacity of traditional security reviews. This mismatch creates blind spots where malicious payloads, insecure libraries, or subtle logic errors can slip into production, amplifying the attack surface for threat actors.

ProjectDiscovery’s survey of 200 security professionals across North America and Western Europe highlights the growing unease. Only 38% of respondents feel they can keep pace with AI‑driven code, and nearly 60% report that the task is becoming more difficult. Concerns cluster around three core risks: exposure of corporate secrets (78%), supply‑chain vulnerabilities from unreliable dependencies (73%), and business‑logic flaws that could be weaponized (72%). European participants flagged secret leakage more often than their U.S. peers, reflecting stricter GDPR compliance pressures.

To mitigate these emerging threats, security teams are calling for robust governance mechanisms before AI tools are integrated. Required controls include immutable audit trails, role‑based access restrictions, and automated verification of generated code against known security baselines. Organizations that embed these safeguards early can balance the speed advantages of AI with the rigor of traditional security practices, preserving both innovation momentum and regulatory compliance. As AI coding matures, the industry will likely see standardized frameworks and vendor certifications that address these trust gaps.