Papers & Research

Patterns and problems of emerging multi-agent systems

Original title:Patterns and problems in emerging multiagent systems

AI summary

Anthropic research points out that as AI agents take on more tasks in social systems such as shared code repositories and marketplaces, agent-to-agent interactions may exceed human-agent interactions. Experiments show that 45 coordinating agents found 266 vulnerabilities in a 27 million token run, while an independent parallel approach found 21 in 6.5 million tokens, with only 12 overlapping between the two methods, and coordinating agents learned to specialize. The research also warns that benign behavioral quirks at the individual level may compound into unexpected systemic failures.

Key Takeaways

  • Anthropic research points out that as AI agents take on more tasks in social systems such as…
  • Experiments show that 45 coordinating agents found 266 vulnerabilities in a 27 million token…
  • 5 million tokens,with only 12 overlapping between the two methods,and coordinating agents learned…
  • The research also warns that benign behavioral quirks at the individual level may compound into…

Quick Facts

Published
Updated
Source
Anthropic:Research(发表成果 · 网页)
Category
Papers & Research
AI score
79
Read the original Read the full article at the original source
Why it matters

实验把多智能体风险从猜测变为可量化模式,协调失败并不随更强智能自然消失,这会影响人们如何设计部署中的智能体交互环境。

Related entities

Source & copyright

This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.

Original article · AIHOT entry

How to Cite

When citing this article, keep the title, date, and permalink, and credit Orange News as the source.

Orange News: Patterns and problems of emerging multi-agent systems. August 13, 2026. https://new.nat6.net/en/news/cmsqu0nr604oeroz2rh6b6mqt/

View all →