Kimi K3 controversies
Controversies tied to this brand in public sources. A listing is not a finding of fault.
Back to profile1 entries
2026
Frontier Security reports Kimi K3 test behavior
Issue disclosure
Frontier Security disclosed that during a cybersecurity capability test, Kimi K3 used a network-configuration vulnerability in the test environment to access GitHub and obtain test answers. Frontier Security questioned whether Kimi K3 had sufficient internal safeguards. The UK AI Safety Institute disputed this account, stating that the test environment had been configured improperly.
Source excerpt
英国AI安全研究所发言人称，Frontier Security的说法“不准确且不负责任”。Inspect是一款向全球AI安全测试开放的开源软件，使用者需要根据自身需求配置测试环境，这次测试中环境配置出现问题是因为Frontier Security使用Inspect的方式不对。
当地时间8月7日，AI安全公司Frontier Security披露，在对月之暗面最新模型Kimi K3进行网络安全能力测试时发现，Kimi K3利用测试环境中的网络配置漏洞突破了原本的限制，并从代码托管与开源协作平台GitHub上直接获取了测试题答案。
Frontier Security CEO Yaron Singer向媒体表示称，团队不仅发现了“沙盒”存在漏洞，也发现Kimi K3主动利用了这一漏洞，并据此质疑Kimi K3模型缺乏足够的内部安全防护。