AI privacy policies
Do they train on your prompts/replies and risk bots that repeating your text to 3rd parties?
--------- API ---- Web UI
Openai No ---- Yes? Maybe? If free or <=$20 account
Google No ---- No
Mistral No --- Yes unless you have a paid account
Anthropic No --- No
AWs Bedrock No --- No
Exceptions, most will train moderation models on your data, which shouldn't have a "text recall" risk
If you thumb up or down a response you've opt-ed in for base model training now