[2308.03825] "Do Anything Now": Characterizing and Evaluating In-The-Wild Jailbreak Prompts on Large Language Models - discu.eu

Reddit

[R] Jailbreak Prompts and LLM Safety https://arxiv.org/abs/2308.03825 4 comments 12/8/2023 machinelearning

Linking pages

Would you like to stay up to date with Computer science? Checkout Computer science Weekly.

Related searches:

Search whole site: site:arxiv.org

Search title: [2308.03825] "Do Anything Now": Characterizing and Evaluating In-The-Wild Jailbreak Prompts on Large Language Models

See how to search.

Submit link to: