Anthropic Study Finds Giving Everyone Their Own AI Agent Backfires

Anthropic's alignment team published a study "Worse Together" showing that assigning each team member an individual AI agent performs worse than having the whole team share one agent, tested across four simulated multi-user environments with frontier models from Anthropic, OpenAI, and Alibaba.

2026-11-15 ~ 2026-11-15 · 2 related posts

• Anthropic study: one agent per person performs worse than a single shared agent — vista8 · 2026-11-15
• Anthropic's MAMUBench: per-user agents coordinate worse than one shared agent — vista8 · 2026-11-15