Large language models are transforming software development by making it easier to write and connect code, but they also introduce serious security risks. Vulnerabilities like LLM command injection, SSRF, and insecure outputs mirror traditional web flaws while creating new attack vectors unique to AI-driven apps.

In this episode, Dan Murphy and Ryan Bergquist discuss how LLM-powered applications can be manipulated into leaking data, executing malicious commands, or wasting costly tokens. They also explain how Invicti’s scanning technology detects and validates these risks, helping organizations protect against the growing challenges of LLM security.