Listen to Jeremiah Salamon, head of OWASP Boston, as he discusses why #PromptInjection remains at the top of the #OWASP Top 10 for LLMs list in 2025.
This is just a small excerpt from a conversation between Jeremiah and Lee Weiner, TrojAI's CEO, that explores the evolving landscape of LLM security and the key changes to the 2025 OWASP Top 10 for LLMs.
Don’t miss this opportunity to get expert insights on the latest threats and how to protect your enterprise from them. Watch the full conversation now! 👉 https://lnkd.in/eBseft4e#TrojAI#OWASPTop10#GenAI#LLMs#cybersecurity
Injection in general, I think the reason why they stayed at number one is because they're starting to not just look at this as just pure input injection, but now there's injection from other sources, there's direct and indirect injection into generative AI. It's not just the individuals putting things into props, but it's the AI solutions are now going out, reaching out to the Internet, reaching out to 3rd party applications and things like that. And that's causing injection issues. And I think that that's why this remains at the number one and will continue to be a problem.
