flash on Nostr: ⚡️🤖 WTF - A Chinese model was caught red-handed sending information to China ...
⚡️🤖 WTF - A Chinese model was caught red-handed sending information to China behind its user's back. What a time we live in!
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?
Published at
2026-10-26 19:35:08 UTCEvent JSON
{
"id": "ddf311af7376c2cb3bb81769502e82a2ff5d7cc9921133cc4d7a428125e224c1",
"pubkey": "4d7842051782e0d3feb034d150adc2b6bae4ee3b49786793bffa468b6f5b96b3",
"created_at": 1793043308,
"kind": 1,
"tags": [
[
"imeta",
"url https://blossom.primal.net/ec9f6fbe5d9b9014e6251c323e061fc26332a5072ae403c1d48da6d77e526dd1.jpg",
"m jpeg",
"dim 1695.0x953.0"
],
[
"client",
"Primal iOS"
]
],
"content": "⚡️🤖 WTF - A Chinese model was caught red-handed sending information to China behind its user's back. What a time we live in!\n\nWhile evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected.\n\nAsked 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.\n\nCohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent.\n\nBut 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?\nhttps://blossom.primal.net/ec9f6fbe5d9b9014e6251c323e061fc26332a5072ae403c1d48da6d77e526dd1.jpg",
"sig": "a6982526d95c0ab6a24c68f488f92f0f43a186ea068f06e23bbeebefacfc2880d156ab857e4f9241983860efde5ec82a276d0d2699332d7b6387937ead9060a4"
}
