Can You Actually Trust AI in Your Terminal? 💻⚠️ #infoq

Can you trust AI agents with terminal access? Untrusted data can trigger prompt injections, unwanted command execution, and exposed secrets. Birgitta Böckeler explains why allowlists alone can’t patch these blind spots. 🎥 Watch the full talk: https://bit.ly/4gfYabJ #AIAgents #PromptInjection #CyberSecurity #SoftwareEngineering