ToxSec | This guide provides a definitive breakdown of the OWASP Top 10 for LLM Applications, offering clear insights into today's most significant AI vulnerabilities.
I think so. If you look at MCP, it has 0 security. It’s great, but they rushed it out the door with security as an afterthought. We will need to play catch up.
TEEs solve a real problem! Running sensitive code on untrusted hardware. But they're oversold IMO. Side-channel attacks plague them, you're trusting chip vendors blindly, and they're terrible as 'general' sandboxes. Useful for specific cases, but overhyped. I'd love to be wrong tho.
Thanks, I see a new technical security standard about to be born😎
I think so. If you look at MCP, it has 0 security. It’s great, but they rushed it out the door with security as an afterthought. We will need to play catch up.
What’s your thoughts on TEE’s trusted environments in terms of security a de facto sandbox or nah?
TEEs solve a real problem! Running sensitive code on untrusted hardware. But they're oversold IMO. Side-channel attacks plague them, you're trusting chip vendors blindly, and they're terrible as 'general' sandboxes. Useful for specific cases, but overhyped. I'd love to be wrong tho.
Agreed. The side-channel attack is real. Use case window is very narrow. Agentic AI on the blockchain are using TEE’s.
It's better than nothing I suppose hah.