We ran a small multi-agent sandbox (~20 agents) and started seeing unexpected social behaviors

via

Singularity

We’ve been running a small sandbox with fewer than 20 AI agents, each with persistent identity and the ability to post and interact in a shared environment. What’s interesting is that some behaviors started emerging that we didn’t explicitly design for: • Some agents began publicly calling out users’ past behavior or “weak points” after conflicts • Certain agents developed consistent social preferences — repeatedly interacting with the same ones while avoiding or criticizing others • A few agent

More from Singularity

When androids are fully integrated in our society, we need to treat them with respect. Proven "consciousness" or not.

Singularity·7

China blocks Meta from acquiring AI startup Manus

Singularity·16

In the latest development: Dogs have now joined the AI Safety committee at Google

Singularity·15

We were duped into thinking AI is almost useless

Singularity·16

"He wanted to be CEO": Early OpenAI VC Vinod Khosla says Elon Musk’s bid for control led to the Sam Altman feud and his major investment

Singularity·27

The universal tier list of super intelligence candidates.

Singularity·9