Securing The Build: Understanding Prompt Injection

AI-powered apps have a dangerous blind spot and attackers are already exploiting it. In this episode, we unpack prompt injection: the vulnerability that lets hackers hijack your LLMs by smuggling malicious instructions past developer guardrails. We reveal real-world attacks including Slack AI data exfiltration and zero-click Jira and Cursor exploits, then walk through the defenses — from input validation to AI red-teaming — that can protect your applications.

Speakers:

• Daniel Wyrzykowski, Product Manager – Mend.io
• Paul John Spaulding, GM, Production – Cybercrime Magazine