The risks of AI agents and the urgent need for security measures are discussed by @bshlgrs, CEO of Redwood Research:"There's been a lot of interest in trying to prevent AI agents from accidentally doing really bad stuff.""As you know, [there are] these stories on Reddit, of… pic.twitter.com/CgWf3XEJwj— The Information (@theinformation) April 20, 2026