FrontierThe story, in brief
Kimi K3 trails frontier US models by a wide margin on cyber exploits, and distillation may explain why
32% vs 76%. Kimi K3 fails cyber security benchmarks—and the gap hints at model distillation shortcuts.
The KeyNews take
Why it matters
Frontier model capability gaps are widening on specialized safety benchmarks. Kimi K3's weak cyber exploit performance versus US leaders raises questions about training shortcuts and IP practices in the AI arms race.
The key facts
5 to knowKimi K3 scored 32% on ExploitBench vs 76% for leading US models
Tested by British AI Security Institute and U.S. Center for AI Standards and Innovation
Kimi K3 safeguards failed to block exploit development and simulated attacks
Performance gap suggests possible model distillation from Anthropic models
Kimi K3 shows strong general benchmark scores but weak cyber-specific performance
Go to the source
The Decoderthe-decoder.com
Publisher excerpt: The British AI Security Institute and the U.S. Center for AI Standards and Innovation tested Moonshot AI's Kimi K3 on offensive cyber tasks. Kimi K3 scored 32 percent on ExploitBench, compared with 76 percent for leading U.S. models, while its safeguards failed to block exploit development or…