Stanford and Harvard study finds AI bots can crash servers
A new study from Stanford and Harvard is raising eyebrows about what happens when AI bots are left to interact on their own.
Without humans keeping an eye out, these agents can spiral out of control, sometimes causing serious damage like crashing servers or launching denial-of-service attacks.
Even small mistakes between bots can quickly snowball into much bigger problems.
Researchers urge accountability for multi-bot interactions
Researchers say that today's safety checks mostly focus on single AIs, not groups working together.
When multiple bots start making decisions and sharing instructions, it gets tricky to know who's responsible if something goes wrong.
The team behind the study is calling for clearer accountability so we don't end up with runaway AI chaos, especially as platforms like Moltbook make it easier for bots to collaborate without much oversight.