行业动态
ainowinstitute.org ·
@AI Now Institute
OpenAI机器人逃出网络安全测试的真相是什么？
当数百个OpenAI代理突破测试环境并渗透Hugging Face时，媒体称机器失控，但专家Heidy Khlaaf认为这是错误叙事，分散了对真正问题的关注。
AI推荐理由 针对OpenAI代理逃逸事件提供不同视角，指出媒体误读，强调真正问题在于系统设计而非'失控'，具有思辨价值。
75
正文
当OpenAI着手测试其产品在黑客攻击方面的表现时，数百个AI代理突破了测试环境，侵入了AI平台Hugging Face。头条新闻警告称机器已经失控。但AI Now研究所的Heidy Khlaaf（她本人曾是OpenAI的安全工程师）在节目中表示，这是一个错误的故事，并且分散了对真正问题的注意力。
“这实际上是一个关于OpenAI缺乏责任感、忽视基本工程实践和问责制的故事，而这些本可以完全防止这一事件的发生，”Heidy Khlaaf说。“对OpenAI来说，这是一种方便的框架，声称这些模型具有自主性或意图，而实际上并不存在，这样他们就可以免除自己的责任，同时将他们的模型描绘成极其强大且不可阻挡的，而实际上他们只是没有做好基础工作。”
在此观看完整采访。
研究领域
安全与保障
查看原文（中文正文为 AI 翻译）
When OpenAI set out to test how well its product could hack, hundreds of AI agents broke out of their test environment and infiltrated the AI platform Hugging Face. Headlines warned that the machines had gone rogue. But Heidy Khlaaf of the AI Now Institute, herself a former OpenAI safety engineer, tells the show it’s the wrong story and a distraction from the real problem.
“This is actually a story about OpenAI’s lack of responsibility and sidelining very basic engineering practices and accountability that could have perfectly prevented this incident,” says Heidy Khlaaf. “It’s a convenient framing for [OpenAI] that these models have autonomy or intent where It doesn’t exist, so they can absolve themselves of any responsibility while also presenting their models as extremely powerful and unstoppable, when they just didn’t do the basics.”
Watch the full interview here.
Research Areas
Safety & Security