Novel AI risks emerge when agents interact. Risks reflect fundamental flaws in the design of agentic software. Responsibility lies with developers to address fundamental flaws. An increasing body of ...