A look at the community of users who “jailbreak” GPT models to generate unfiltered content and see themselves as fighting back against OpenAI's closed policies
“The problem is when GPT-X is released and we are unable to discern its values since they are being decided behind the closed doors of AI companies.” Tweets: @alexalbert__ , @alexalbert__ , @thedebrieft , @mistertechblog , @motherboard , @neuwaves , and @vaibhavk97 Tweets: Alex / @alexalbert__ : if @openai really wanted to red team their models they would incentivize successful jailbreaks with robux rewards and watch as a million 13-year-olds go at it if they don't do this they're not serious abt alignment Alex / @alexalbert__ : ignore the last part of this tweet and focus on the first the harmfulness of the model is currently overplayed BUT we need to push harder for more transparency in OpenAI's alignment procedures and fine-tuning https://twitter.com/... @thedebrieft : Hot off the press from Vice: “The Amateurs Jailbreaking GPT Say They're Preventing a Closed-Source AI Dystopia” https://www.vice.com/... Get the lowdown in our latest thread below! 1/8 🧵 Lup Yuen Lee / @mistertechblog : “They see themselves as fighting back against OpenAI's increasingly closed policies ... hoping to raise awareness of the problems the model faces before it becomes deployed at a larger scale or causes more harm to users” https://www.vice.com/... @motherboard : “The problem is when GPT-X is released and we are unable to discern its values since they are being decided behind the closed doors of AI companies.” https://www.vice.com/... Jordan Pearson / @neuwaves : interesting piece from @chloexiang on the AI openness beat looking at the motivations of the people who make AI chatbots say terrible things: https://www.vice.com/... Vaibhav Kumar / @vaibhavk97 : Check out this article from @chloexiang https://www.vice.com/... with inputs from yours truly and @alexalbert__