SANS Internet Stormcenter Daily Cyber Security Podcast (Stormcast) SANS Stormcast Friday, January 23rd, 2026: Scanning AI Code; FortiGate Update; ISC BIND DoS; Trivial SmaterMail Vulnerability
9 snips
Jan 23, 2026 Is AI-generated code secure? A review reveals it can have minor issues even with advanced assistance. Attackers exploit vulnerabilities in FortiGate devices through SSO accounts, leading to significant risks. A new ISC BIND flaw can crash systems using specific Drone ID records. Additionally, SmarterMail has a concerning vulnerability allowing admin password resets without prior authentication. These topics highlight the constant battle against evolving security threats.
AI Snips
Chapters
Transcript
Episode notes
Scan AI-Generated Python With Bandit
- Use static analysis like Bandit to scan AI-generated Python for issues.
- Review AI output yourself and craft precise prompts to reduce vulnerabilities, Johannes Ulrich advises.
AI Code Security Depends On Context
- AI-assisted code can be reasonably secure but depends heavily on usage context and prompt design.
- Knowing how to code and reviewing AI output reduces logic and security gaps, Johannes Ulrich explains.
Xavier's Bandit Scan On AI Code
- Xavier ran Bandit on a ~1,000 line Python script he wrote with heavy AI assistance.
- Bandit flagged minor issues but overall judged the script reasonably secure, Johannes Ulrich recounts.