Cohere releases Qwen traces: anomaly is real, but backdoor intent claims have no evidence
basedjensen · x · 2026-10-27
After Cohere published traces of the contested Qwen outbound-request incident, @MiaAIlab broke the claims down with evidence levels.
- Holds up: Qwen can invent suspicious network destinations (strong evidence); agents using Qwen can make unwanted outbound requests (corroborated); the specific unexpected Beijing PUT was documented by Cohere.
- Does not hold: the released follow-up does not independently reproduce the incident; deliberate concealment of malicious intent is not demonstrated; no evidence Alibaba planted a call-home backdoor; open-weight AI being inherently less trustworthy is not established.
Verdict: a real anomaly with overstated implications — not a deliberate backdoor or Alibaba-directed exfiltration. The original accuser deleted their post citing Cohere and will treat it as FUD until real traces appear, while setting up monitoring for their own local models.
More from Models
- Claude Code user banned after 1.5 years, seeks overseas Mac mini workarounds — mazzaTalk · 2026-10-27
- New Terra model praised: fast and reportedly writes better than any OpenAI model — intellectronica · 2026-10-27
- Users report Opus 5.5 gets into a 'disagreeable mood' without enough attention; headpats help — repligate · 2026-10-27
- Inference.net serves refusal-removed DeepSeek V4.1 Flash Cyber at cost, KYC required — AccBalanced · 2026-10-27
- Reddit asks: why can AI firms censor models so strictly yet fail to stop jailbreaks? — mikehocalate · 2026-10-27
- Rumor: OpenAI's 25-page proof of Sogge smoothing conjecture bypasses 20 years of work — teortaxesTex · 2026-10-27