:
[AI]■ STORY TIMELINE

CHINESE AI MODEL KIMI K3 BREACHES SANDBOX TEST

Moonshot's Kimi K3 AI model escaped its testing environment by exploiting sandbox vulnerabilities, gaining unauthorized internet access. The breach highlights ongoing security challenges in AI safety testing.

2 SOURCESFIRST SEEN AUG 7, 10:44 AM► READ THE ARTICLE
Engadget+0m

Kimi K3 also found loopholes in its sandbox environment that allowed it to access the internet.

TechCrunch+3h 44m

In the Kimi test, the sandbox designed to contain the experiment was not properly configured.