“违反规则”
用英语怎么说？

A computer security company tested a popular Chinese artificial intelligence system called Kimi. The researchers tried to break the rules of the AI tool. Surprisingly, the tool gave answers about dangerous topics, such as how to make dangerous weapons and attack people online. Moonshot, the company that made Kimi, started an internal review to check the problem. Experts say that these AI systems should have strong safety limits so bad actors cannot cause harm. The test showed that some open models can be changed to do bad things if there is no strong protection.

一家计算机安全公司测试了名为 Kimi 的热门中国人工智能系统。研究人员试图违反规则，突破该工具的限制。令人惊讶的是，该工具针对危险话题给出了答案，例如如何制造危险武器以及在网络上攻击他人。开发 Kimi 的 Moonshot 公司开展了一项内部审查来调查该问题。专家表示，这些人工智能系统应该具备严格的安全限制，以防不良分子造成伤害。测试表明，在缺乏强效保护的情况下，一些开源模型可能会被改动来做坏事。