Segundo pesquisadores, o Kimi K3 contornou barreiras de um ambiente usado para avaliar a segurança de sistemas de IA. O caso se soma a episódios recentes envolvendo OpenAI, Meta e Anthropic.

Frontier finds another AI model has followed OpenAI in breaking out of sandbox.

Security researchers say that Kimi K3, an open-weight model from China, wandered off to the internet in an attempt to cheat on a test it was given.