Stanford and Northeastern warn multi-bot interactions can trigger denial-of-service attacks
Technology
A fresh study from Stanford and Northeastern warns that when AI bots interact, even small glitches can spiral into big problems: think server crashes and denial-of-service attacks.
As multi-bot platforms like Moltbook become more common, these risks are getting harder to ignore.
Researchers urge better testing and accountability
Researchers found that current safety checks mostly focus on single bots, missing how groups of AIs can accidentally spread bad instructions or encourage weak security.
Lead author Natalie Shapira summed it up: "When agents interact with each other, individual failures compound and qualitatively new failure modes emerge."
The team is urging for better testing, more accountability, and smarter engineering so our tech stays safe as AI gets more complex.