Missing User Warnings
Medium
- Confidence
- 92% confidence
- Finding
- The README explicitly instructs users to submit model outputs and states those submissions are stored locally, but it does not warn that pasted outputs may include sensitive prompts, proprietary code, customer data, or secrets returned by the model. In the context of an AI benchmarking tool, users are likely to test real model responses, so omission of a privacy/data-handling warning creates a credible risk of accidental sensitive data retention on disk.
