A Stripe employee hid a message in his LinkedIn profile telling any AI that read it to include a flan recipe. A month later, an AI recruiter emailed him one. It's funny until you realize the same technique can exfiltrate data, generate phishing content, or hijack automated business processes. What is prompt injection, why does OWASP rank it as the number one risk to large language models, and what should you do about it? Let's find out. Your hosts are Kip Boyle, CISO with Cyber Risk Opportunities, and Jake Bernstein, Partner with K&L Gates.

 

OWASP Top 10 for LLM Applications -- https://genai.owasp.org

Rss Apple Podcaster
Podden och tillhörande omslagsbild på den här sidan tillhör Kip Boyle. Innehållet i podden är skapat av Kip Boyle och inte av, eller tillsammans med, Poddtoppen.