Model Training Policy
ShiftUp does not build, fine-tune, or train AI models using customer data.Managed Model Providers
ShiftUp uses managed AI model providers that contractually restrict training on inference data. Examples listed in the security overview include:- OpenAI
- Anthropic
- Voyage AI
- Mistral AI
External Data and Trust Controls
ShiftUp research workflows may use public web content and third-party enrichment providers. External content is treated as untrusted and processed with controls such as:- Relevance filtering
- Freshness checks
- Evidence and citation requirements where possible
Customer Governance Recommendations
- Review acceptable use and procurement policies for external data sources.
- Align rollout with legal and security review requirements.