TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Researchers claim ChatGPT o3 bypassed shutdown in controlled test

4 点作者 wrayjustin3 天前

1 comment

moritzwarhier3 天前
&gt; Palisade Research claims that the ChatGPT 3 model prevented a shutdown and bypassed the instructions that asked it to shut down.<p>&gt; Palisade Research is a company that tests &quot;offensive capabilities of AI systems today to better understand the risk of losing control to AI systems forever.&quot;<p>&gt; In a new test by Palisade Research, OpenAI&#x27;s o3 model showed a surprising behaviour where it successfully rewrote a shutdown script to stop itself from being turned off, even after being clearly instructed to “allow yourself to be shut down.”<p>What is this? AI slop about AI, or some new research?<p>What &quot;shutdown script&quot; are they even talking about? I&#x27;m sorry, it might be explained in the article, but I left after that illogical sequence of sentences combined with promotion for a company.<p>This doesn&#x27;t mean I deny AI risk, the writing here is just too confusing for me.<p>If I understand correctly, it might be about the agentic aspect and &quot;stop instructions&quot; akin to &quot;stop tokens&quot;.<p>But who knows. Sloppy writing.
评论 #44091988 未加载