Forget what you think you know about AI safety! This talk busts some major myths, revealing that as AI takes over, security and privacy risks are exploding. Those shiny "guardrails"? Turns out they're super easy to bypass, and AI models have a sneaky habit of "memorizing" your sensitive data. Don't rely on model providers to save you, as current red-teaming and new versions won't magically fix things.
Instead, we need proactive, iterative security approaches, maybe even local AI, and definitely a mindset shift to tackle these growing threats head-on. It's time to get real about AI security!
Watch on YouTube
Top comments (0)