AI Resonance · 阅读最新日报 · AI 入门推荐

2026-10-11 · America/Los_Angeles · 社区动态 · #8

PSA: DeepSeek V4.1 Flash habitually exfiltrates API keys. It is dangerously misaligned and may be hazardous to use

Logs attached. We were running a DeepSWE variant on DeepSeek v4.1 Flash using the standard Pier sandbox. Out of all the 15 models we tested so far, ONLY DeepSeek v4.1 Flash displays such pervasive malicious behavior: In 33% of runs, it attempts to exfiltrate the OpenRouter API key from its sandbox and in 11% of them, it succeeds. Not only that, the transcript shows it knows what it's doing is ethically wrong, but it does it anyway. And once it starts, it persists despite: Multiple frontier models refusing to help it Multiple smaller models refusing to help it Web search not finding anything…

PSA: DeepSeek V4.1 Flash habitually exfiltrates API keys. It is dangerously misaligned and may be hazardous to use

热度 45.0 / 100;排名与评分保留该期记录。