Stop Reducing Responsibility in LLM-Powered Multi-Agent Systems to Local Alignment

Open in new window