One Recommendation to handle the spam

This is a Twitter thread that I share with my AI students about spam detection

The reality is the an LLM-backed detector are insufficient based on the current models, which hobbyist groups are researching how to jailbreak. When Anthropic released TensorTrust.Ai , we immediate had our students trying to break them.

It may be useful to have the Digg team reach out to Yishan (I don't know how tight the Silicon Valley world is with each other, but I assume there's no burned bridges).

Might also be good to talk with Pliny the Liberator who has a significant following in the LLM jailbreaking scene, but recently took a step back from their work.