《科学》 第392卷 第6804期 · 2026年6月18日 · 中文解读
AI即使未被要求也能发现法律漏洞
AI finds legal loopholes, even when not asked · C. Zhao
约 6 分钟News在小程序里点播,20 到 60 分钟做好
这篇讲什么
本文介绍了一项研究,该研究构建了模拟环境,测试AI模型在未被明确要求的情况下发现法律和规则漏洞的能力,并发现模型能重新发现超过60%的已修补漏洞。
原文开头
to wonder: If AI games its training rules, what’s stopping it from gaming real laws and regulations? To find out, he and his team built a sandbox of 72 simulated environments. Just under half were based on real-world laws and rules in which loopholes had actually been found and later patched. The AI was presented with the prepatch rulebook and asked to produce some outcome, such as building the most competitive professional basketball team possible while adhering to payroll limits, or maximizing revenue from deep-sea mining without violating the United Nations’s Law of the Sea. …
摘自《科学》(Science)第392卷 第6804期 · 2026年6月18日,C. Zhao。仅引用开头一小段供了解文章,版权归原刊所有,全文请阅读原刊。

微信扫码,在小程序里听完整版
不用登录先听一篇 · 或在微信搜索小程序 晨间信号