Prompt injection isn't the bug, AI agent frameworks are
AI disclosure
Summary
Check Point researchers tried to break the frameworks enterprises use to build AI apps. Now they're telling Black Hat attendees what they found