# Where Do My Prompts Go? > A sourced tracker of how 25 AI tools handle user data: training on your data, retention, zero data retention (ZDR), abuse-monitoring exceptions, data residency, SOC 2 / ISO 27001, HIPAA BAAs and self-host options. Every value links to the vendor's own page with a quote and the date it was checked. Last full check: 2026-10-26. 200 fields in total (25 tools x 8 fields). 26 are marked "unverified": no official page stated it clearly, so there is no value. Policies differ by plan; a signed contract overrides public pages. This is not legal advice. ## Data - [data.json](https://rebuildlab.github.io/where-do-my-prompts-go/data.json): full dataset. Each vendor has fields; each field has value, detail, quote, source (URL) and checked (date). - [data.csv](https://rebuildlab.github.io/where-do-my-prompts-go/data.csv): same data, one row per tool, five columns per field (value, detail, quote, source, checked). ## Fields - trains_by_default: Trains on your data by default - retention: Default retention - zdr: Zero data retention - abuse_monitoring: Abuse-monitoring exception - residency: Data residency / regions - certifications: SOC 2 / ISO 27001 - hipaa_baa: HIPAA BAA - self_host: Self-host option ## Pages - [Dashboard](https://rebuildlab.github.io/where-do-my-prompts-go/): status panels, policy event stream and comparison table of all tools - [Method](https://rebuildlab.github.io/where-do-my-prompts-go/about/): how values are sourced and what "unverified" means - [Changelog](https://rebuildlab.github.io/where-do-my-prompts-go/changelog/): confirmed policy changes, checker runs and fetch errors - [Funding](https://rebuildlab.github.io/where-do-my-prompts-go/funding/): who pays for the site (no one yet) and the rules for sponsors and affiliate links ## Tools - [OpenAI API](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/openai-api/): Model APIs; trains by default: no; ZDR: on approval - [Anthropic API (Claude)](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/anthropic-api/): Model APIs; trains by default: no; ZDR: on request - [Google Gemini API](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/gemini-api/): Model APIs; trains by default: depends; ZDR: no (use Vertex AI) - [Mistral AI (Studio / API)](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/mistral/): Model APIs; trains by default: depends; ZDR: on request (PAYG only) - [Cohere](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/cohere/): Model APIs; trains by default: depends; ZDR: on request (enterprise) - [xAI API (Grok)](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/xai/): Model APIs; trains by default: no; ZDR: yes (ZDR-enabled API) - [DeepSeek API](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/deepseek/): Model APIs; trains by default: unverified; ZDR: unverified - [Groq](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/groq/): Model APIs; trains by default: no; ZDR: yes (self-serve) - [Cursor](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/cursor/): Coding tools; trains by default: depends; ZDR: yes (Privacy Mode) - [GitHub Copilot](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/github-copilot/): Coding tools; trains by default: depends; ZDR: partial - [Claude Code](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/claude-code/): Coding tools; trains by default: depends; ZDR: Enterprise, on request - [OpenAI Codex](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/openai-codex/): Coding tools; trains by default: depends; ZDR: unverified - [Windsurf (Devin Desktop)](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/windsurf/): Coding tools; trains by default: depends; ZDR: on opt-out (paid) - [JetBrains AI](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/jetbrains-ai/): Coding tools; trains by default: depends; ZDR: unverified - [OpenRouter](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/openrouter/): Inference hosts & routers; trains by default: depends on provider; ZDR: yes (routing filter) - [Together AI](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/together-ai/): Inference hosts & routers; trains by default: no; ZDR: yes (opt-in, org setting) - [Fireworks AI](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/fireworks-ai/): Inference hosts & routers; trains by default: no; ZDR: default - [Ollama Cloud](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/ollama-cloud/): Inference hosts & routers; trains by default: no; ZDR: default - [AWS Bedrock](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/aws-bedrock/): Inference hosts & routers; trains by default: no; ZDR: default (most models) - [Azure OpenAI (Microsoft Foundry)](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/azure-openai/): Inference hosts & routers; trains by default: no; ZDR: on approval (modified abuse monitoring) - [Langfuse](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/langfuse/): Eval & monitoring; trains by default: no; ZDR: n/a - [LangSmith](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/langsmith/): Eval & monitoring; trains by default: no; ZDR: n/a - [Braintrust](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/braintrust/): Eval & monitoring; trains by default: no; ZDR: n/a (built-in model endpoints are ZDR) - [Arize (AX / Phoenix)](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/arize/): Eval & monitoring; trains by default: unverified; ZDR: n/a - [Helicone](https://rebuildlab.github.io/where-do-my-prompts-go/vendors/helicone/): Eval & monitoring; trains by default: unverified; ZDR: partial (Omit Logs)