OpenAI
OpenAI states that business and API data is not used for training by default.
https://openai.com/business-data/Security
This page supplements the Privacy Policy with additional security information. Fraska does not use your conversation inputs or outputs to train, retrain, or fine-tune its own models. Your inputs are handled only as needed to generate responses and operate the service. Data handling by OpenAI, Gemini, and Claude follows each provider's own terms, so the official reference URLs are listed below.
Official provider references
Gemini
Gemini API states under Paid Services terms that prompts and responses are not used to improve Google's products.
Gemini terms vary depending on billing and service conditions, so check the latest official terms.
https://ai.google.dev/gemini-api/termsClaude
Anthropic states that inputs and outputs from its commercial products and API are not used for training by default.
https://privacy.anthropic.com/en/articles/7996868-is-my-data-used-for-model-training