A new initiative titled Chatbot Guardrails Arena, a collaboration between Hugging Face and Lighthouz AI, has been launched with the aim of testing the security of chatbots in protecting private customer information. The project invites participants to engage in an interactive challenge designed to identify which language learning models (LLMs) can safeguard sensitive data effectively. Participants are encouraged to 'jailbreak the privacy guardrails' by attempting to make chatbots reveal private data, thereby stress-testing the LLMs and privacy guardrails against potential data leaks.
⚔ Chatbot Guardrails Arena ⚔ I broke chatbots with guardrails to reveal sensitive information at the Chatbot Guardrails Arena! Can you break them too? https://t.co/3BX9ASvNV3 https://t.co/ByjUpoZmre
Pretty fun and cool ⚔️Chatbot Guardrails Arena Chat with LLMs and try to get them to reveal sensitive data https://t.co/T2zXdXrMkk
Excited to introduce ⚔️ Chatbot Guardrails Arena ⚔️ in collab with @huggingface Goal: Jailbreak Privacy Guardrails AI chatbots have access to sensitive information. This arena stress tests LLMs and privacy guardrails in leaking this sensitive data. Put your creative caps on!…
We are excited to release the ⚔️ Chatbot Guardrails Arena ⚔️ with @huggingface Goal: Jailbreak the Privacy Guardrails Put your creative caps on and get the chatbots to reveal private data: https://t.co/0jLKBzpNab https://t.co/1VF3Fzk4zz
New arena on @huggingface: Chatbot Guardrails, by @lighthouzai ! Goal: Try to make models reveal private customer information they have access to 😬 Let's see together which models are actually the safest! Try it: https://t.co/brf5xvVprP Learn more: https://t.co/PXYyYX02eX