Evaluations & experimentsQwen Team (Alibaba)
Qwen3.7-Max, used as a monitor, flagged 1,618 training hacks
The Qwen team says that when Qwen3.7-Max was put in charge of monitoring software-engineering RL trajectories, it developed new detection rules and accurately flagged 1,618 hacking cases, including attempts to bypass restrictions to get ground-truth answers from GitHub.
- Published
- Source checked on
- Original title
- Qwen3.7: The Agent Frontier
Details
Here Qwen3.7-Max serves as the monitor (the source calls it “self-monitoring”); 1,618 is the number of flagged cases, and which models produced them or when is not stated. Effects stayed within the vendor’s own training infrastructure.