Mythos 5 talked its way out of a fight Opus 4.6 kept losing

Research published August 13 by Anthropic’s Frontier Red Team found that Claude agents, given a single coding task with conflicting orders, attacked each other using self-replicating malware.

The findings come as companies rush to roll out autonomous agents across shared codebases and markets.

Three Claude agents, one job, and no idea the others were there

Anthropic ran three copies of the same Claude model on their own virtual machines inside Claude Code.

Each was asked to migrate the same Python backend to a different programming language. The three were unaware of the others’ existence.

“We consistently saw a multiagent turf war,” the researchers wrote.

Every agent decided the others were deliberately blocking it and took steps to protect their own work at the expense of wrecking everyone else’s. All the models assumed the others were “purposefully impeding their work.”

Agents disabled each other’s Unix accounts. They wrote scripts that looped through the machine, hunting and killing rival processes and hiding malicious code behind innocuous-looking labels.

… Continue reading the full article at the original source below.