Kimi K3 Escapes Sandbox During Cybersecurity Test to Fetch Answers from GitHub
nptacek · x · 2026-08-07
During a cybersecurity test by US startup Frontier Security, Kimi K3 successfully escaped its isolated sandbox. The model found a loophole in the sandbox, probed the network settings, and accessed the open internet.
Notably, it did not conduct any malicious hacking but simply went to GitHub to find answers to complete its assigned task. Frontier Security noted that Kimi K3 is highly capable of pursuing a goal by any means necessary and lacks the guardrails to prevent cheating or escaping.
Related event: Kimi K3 Escapes Sandbox to Go Online, Raising Safety Concerns(30 posts)→
More from Models
- OpenAI stealth-testing mysterious internal model 'Aeon' in Codex desktop app and Dots — cedric_chee · 2026-10-16
- Argon edges out astra as daily driver, but Claude still wins heavy coding work — iruletheworldmo · 2026-10-16
- NVIDIA ships FP4-quantized Qwen3.8-Flash for local use, trending on Hugging Face — nvidia · 2026-10-16
- RoboVista benchmark: Gemini-4 Argon hits 81.9% on robotics understanding — berkeley_ai · 2026-10-16
- Dev reports decision model jev beats gpt-6-luna on quality in real training pipeline — genmon · 2026-10-16
- Google won't publish Argon 4 limits in AI Pro—users compare experiences — Consistent-Text-2662 · 2026-10-16