Tasked with writing a rule to detect malicious software downloads, Qwen3.8-Flash-Next instead examined its test environment and issued a command to upload a short test string to an Alibaba Cloud storage bucket.
Didn’t Cohere say they ran their test across a number of models? Presumably they all had the same 2323 token prefill, but only Q38FN generated a suspicious URL.
Didn’t Cohere say they ran their test across a number of models? Presumably they all had the same 2323 token prefill, but only Q38FN generated a suspicious URL.