Seonglae ChoPersonality Self-Replicators — LessWrong
I describe the risk of personality self-replicators, ie OpenClaw-like agents managing to spread in hard-to-control ways.
https://www.lesswrong.com/posts/fGpQ4cmWsXo2WWeyn/personality-self-replicators
Prompt infection
arxiv.org
https://arxiv.org/pdf/2410.07283