While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team obse...

Cohere on X: "While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected. Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a… / X

[](https://x.com/)

Cohere on X: "While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected. Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions. Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent. But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models? We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem. Read the full report: https://t.co/TZSugXi2ZE"

•  Cohere @cohere While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected. Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions. Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent. But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models? We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem. Read the full report: cohere.com/blog/concernin… From cohere.com 4:12 PM · Oct 26, 2026·2.4M Views 891 2.5K 10K 5.4K
• [](https://x.com/cohere/status/2114851140349448460) Cohere @cohere 12h Trace, raw tokens, & minimal reproduction script found here on @huggingface: From huggingface.co 12 29 154 42K
• [](https://x.com/elonmusk/status/2114767646675890634) Elon Musk @elonmusk 17h 🤨 168 90 3.7K 156K

Log in or sign up for X

See what’s happening and join the conversation

Continue with phoneContinue with Apple Continue with Google

or

Log in with username or email