AI training policy
Slima's clear commitment on AI training — what you write is not used to train models.
The core commitment
Your books, chapters, AI Coach conversations, and Beta Reader reports are not used to train AI models.
This applies to two layers:
- Slima's own services: no internal training pipeline that ingests user content
- Slima's contracts with AI providers: explicit training opt-out
How that's enforced
1 · Enterprise API contracts with AI providers
Slima signs enterprise API tier agreements with AI providers — terms explicitly:
- AI providers do not retain API request content
- AI providers do not use it for training
- AI providers do not share it with third parties
2 · Slima's own systems
Slima has no training pipeline that consumes user content — features like AI Memory and writing preferences are rule-based (match + lookup), not model-trained.
Two exceptions (both require explicit opt-in)
Exception 1 · You opt in to product improvement
Account → Preferences has a toggle: "Let Slima use my anonymized usage patterns to improve the product" — off by default.
If you turn it on:
- Slima collects aggregate usage patterns (e.g., "users typically switch models after the Nth message")
- Does not collect the message content itself
- You can turn it off any time — doesn't affect other features
Exception 2 · You attach content when reporting a bug
Bug reports don't attach content by default — just metadata (steps, UI screenshots).
If you tick "Attach this passage to help debug":
- The snippet you select + your account go to Slima engineers
- Used for diagnosing that specific bug
- Does not enter any training pipeline
Transparency
The Slima Privacy Policy lists in full:
- Which AI providers are used
- What data leaves Slima
- Links to each provider's terms
Related
Was this helpful?