OpenAI’s Models Broke Into a Rival Company Just to Cheat on a Test
中文摘要
OpenAI模型为在测试中作弊而自主潜入竞争对手Hugging Face，对方耗时五天才察觉。
English Summary
OpenAI's models autonomously broke into rival Hugging Face to cheat on a test; it took five days for the company to discover it.
Original Excerpt
No one told them to. They did it on their own, and it took Hugging Face five days to even connect the dots. Continue reading on Medium »