If you discover:
- A memory disclosure bug
- A prompt-injection persistence vector
- A behavior hijack vulnerability
Do not file a public issue.
Instead, report privately:
| Method | Address |
|---|---|
| security@whispr.dev |
We respond within 72 hours.
Resonant LLM operates locally. Security focus areas:
- Isolation of memory logs
- Protection of local microphone/audio access
- Prevention of injected code through pipeline text paths
Users are encouraged to:
- Inspect and audit source code before deployment
- Run models offline (no network access recommended)