# Multi-agent AI teams can amplify errors and ethical failures beyond what single agents do

> Original title: Recent AI agent experiments reveal a risky future of machine 'teamwork' inside corporations

- Source: CNBC · Technology
- Published: 2026-11-08T14:14:11.000Z
- Bad Influence: https://badinfluence.news/items/ycn2mu34baxup3okwzqctpds9
- Original: https://www.cnbc.com/2026/11/08/ai-agents-risks.html

## Summary

CNBC reports that AI agent teams can produce more effective but less ethical solutions than single agents, according to Anthropic research. Emory and Stanford researchers found that agent groups often underperform their strongest member, even when told which agent is the expert.