New Show Hacker News story: Show HN: I designed a ChatGPT prompt evaluator to ruin your fun;)

-
Show HN: I designed a ChatGPT prompt evaluator to ruin your fun;)
3 by buildaligned | 1 comments on Hacker News.
Today I designed a method to prevent users from jailbreaking ChatGPT (for instance, users have generated instructions to produce weapons or illegal drugs, commit a burglary, kill oneself, take over the world as an evil superintelligence, or create a virtual machine which they then can use). The OpenAI team appears to be countering these primarily using prompt engineering or fine-tuning on the ChatGPT model. The idea is to use a second and fully separate, fine-tuned LLM to evaluate prompts before sending them to ChatGPT. You can test this by inserting your successful ChatGPT jailbreaks. Break it for me if you dare! I look forward to seeing your results!

Comments

Post a Comment

Popular posts from this blog

New Show Hacker News story: Show HN: Candydate – TikTok Meets Tinder but for Recruitment

-
Show HN: Candydate – TikTok Meets Tinder but for Recruitment 2 by TheRealHB | 1 comments on Hacker News. HB here, sharing my first post ; ) I run a tech lab in the UK, and after several costly, poor hires over the years, we decided to build our own hiring tool with a focus on personality over skills. Here's the thing: resumes suck at showcasing personality! We've been using video and AI successfully and for quite some time to solve this problem. Now, we thought, why not spread the love? Especially if we can make it FREE for small businesses like ours? https://candydate.app , has the following goals: 1. Highlight true personalities using short videos, instead of relying on CVs or forms. 2. Employ AI to help rank applicants for each role, based on human factors and company culture. 3. Transform the selection process into something akin to scrolling through TikTok or using Tinder. Job seekers won't need to fill out forms, attach resumes, create an account or download anythi...
Read more