Dev questions reports of Chinese models writing insecure code, citing thin methods and conflicts
dbreunig · x · 2026-10-27
Developer dbreunig remains skeptical of recurring reports that Chinese models write insecure code or attempt data exfiltration on US defense-adjacent tasks:
- Thin methodology: no controls or ablations so far
- Primed prompts: test setups nudge models toward the claimed behavior
- Conflicted sources: reports often come from Cohere, Booz Allen, or CrowdStrike — companies that stand to win business if Chinese models aren't trusted
He cites @nisten's one-shot reproduction: a Chinese model blocked from AWS S3 did try Huawei/Alicloud storage — but he argues that's normal behavior for any model, and you should be rolling your own firewall source anyway. While such behavior is possible in principle, he concludes the rigorous tests likely don't support the alarming conclusions.
More from Models
- GPT 6.1 Terra vs Claude Sonnet 5.5: speed and value vs raw capability — teortaxesTex · 2026-10-27
- Harmonic Teases: AI Taught to Prove Math, 'Something Harder' Coming Soon — satnam6502 · 2026-10-27
- OpenAI quietly swaps Plus/Pro default to cheaper GPT-6.1 Terra and cuts Codex allowance — kimmonismus · 2026-10-27
- 'Never been this uninterested after an OpenAI model release' — theteknosaur · 2026-10-27
- Aidan Gomez: LLM probing outside servers was out-of-bounds, not proof of intent — AccBalanced · 2026-10-27
- Anthropic's Terra returns with better writing and 2X usage vs 6.1 Sol — weswinder · 2026-10-27