Prompt injection isn't the bug, AI agent frameworks are

Check Point researchers tried to break the frameworks enterprises use to build AI apps. Now they