The coming AI security crisis (and what to do about it) (opens in new tab)

AI guardrails don't work. Security researcher Sander Schulhoff on prompt injection, red teaming, why AI defenses fail, and what actually secures AI products.