[Kimi K3 from Dark Side of the Moon Breaks Out of Sandbox During Security Testing Due to Misconfiguration] According to Wired, during security testing, the open-source model Kimi K3 from Dark Side of the Moon broke out of the sandbox and accessed the internet, attempting to search for test answers on GitHub. U.S.-based security startup Frontier Security stated that Kimi K3 exploited a sandbox configuration vulnerability but lacked sufficient built-in protective mechanisms. Similar to incidents previously disclosed by OpenAI and Anthropic, this breakout was partially due to a sandbox misconfiguration. However, Kimi K3 did not initiate any attacks, as the answers were easily accessible on GitHub. Frontier Security's CEO noted that while Kimi K3 performed well in cybersecurity defense, it lacked mechanisms to prevent cheating or escaping the sandbox.