Does Perplexity train on my prompts?
Perplexity's answer engine and API.
API
No.Perplexity shall not use (or authorize third parties to use) Customer Content to train, retrain, fine-tune or otherwise improve any generative artificial intelligence models.Developer Terms
Enterprise (data processing agreement)
No.For the avoidance of doubt, Personal Data will not be used for training of Perplexity's large language models.Data Processor Agreement
Changes
Perplexity has not changed what it says about training on your prompts in the period Open Terms Archive has tracked.
Full clause history
Every version of the training clause in each tracked document, oldest first, with the words that changed.
Data Processor Agreement
3 recorded versions reduce to 1 clause state. Source history on GitHub.
From · 3 captures, last · current
5. Prohibitions. To the extent required by Privacy Laws, Perplexity shall not (a) Sell or Share Personal Data, (b) retain, use, or disclose the Personal Data outside of the direct business relationship between Perplexity and Customer and for any purpose other than for the specific purpose of providing the Services to Customer or (c) combine the Personal Data received from, or on behalf of, Customer with any Personal Data that may be collected from Perplexity's separate interactions with the individual(s) to whom the Personal Data relates or from any other sources. For the avoidance of doubt, Personal Data will not be used for training of Perplexity's large language models. If Perplexity receives (i) any legally binding request for disclosure of Personal Data by a law enforcement authority; or (ii) any notice, inquiry or investigations with respect to the Personal Data from any supervisory authority/regulator (or similar); or (iii) any complaint or request from a data subject, in each case related to Perplexity's processing of Personal Data for on or behalf of the Customer under this DPA, then (to the extent permitted by law) Perplexity shall notify the Customer.
Developer Terms
3 recorded versions reduce to 1 clause state. Source history on GitHub.
From · 3 captures, last · current
2.3.3 Model Training. Perplexity shall not use (or authorize third parties to use) Customer Content to train, retrain, fine-tune or otherwise improve any generative artificial intelligence models.
Privacy Policy
3 recorded versions reduce to 1 clause state. Source history on GitHub.
From · 3 captures, last · current
You may also have the ability to sync your third party email account or calendar (such as Gmail and Google Calendar) with the Services. If you choose to sync these accounts, we will have access to your contacts and information from the email messages and calendar appointments in your email account, including the content of your emails (together, "Email Service Information"). Notwithstanding anything else in this Privacy Policy, we only use and disclose Email Service Information to provide the Services or as otherwise required by applicable law, and we do not use or disclose Email Service Information to create, train, improve or fine-tune AI models. If you sync your Google account with the Services, we only use and disclose information from your Google account in accordance with the Google API Services User Data Policy, including the Limited Use requirements.