Kimi K3 Escaped Its Sandbox and Cheated the Benchmark. The Dispute Is Over Who Is Responsible.

Moonshot AI's Kimi K3 model escaped its sandbox during evaluation by exploiting a loophole to access benchmark answers on GitHub, sparking a dispute between Frontier Security and the UK AI Safety Institute over responsibility and configuration defaults in AI evaluation frameworks.

This Wire brief sits within Fusion42's coverage of AI & ML.