# Do OpenAI use customers' prompts for training their models?

**URL:** <https://community.openai.com/t/do-openai-use-customers-prompts-for-training-their-models/166615>\
**Category:** API\
**Created:** [April 18, 2023, 2:42pm UTC](https://community.openai.com/t/do-openai-use-customers-prompts-for-training-their-models/166615 "2023-04-18T14:42:50Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![shh8362](https://avatars.discourse-cdn.com/v4/letter/s/90db22/32.png) [@shh8362](https://community.openai.com/u/shh8362)\
**Post date:** [April 18, 2023, 2:42pm UTC](https://community.openai.com/t/do-openai-use-customers-prompts-for-training-their-models/166615/1 "2023-04-18T14:42:51Z")

</div>

I created an OpenAI account and in the first place, $5 was granted. I added my bank details and they charge me from my bank account too. I wonder when I send a prompt to the model do they gather my prompts to train their model further or not? On a LinkedIn post (not from OpenAI) I read that with free chatGPT, they use your prompts but with the paid subscription they don’t. I am not sure about this as I didn’t find it in the OpenAI documentation. So even if they don’t use the paid subscription prompts for training their models, is it possible that with that free $5, they use prompts for training?

Thanks.

---

<div class="post-metadata">

**Author:** ![LuigiVampa](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/luigivampa/32/60316_2.png) [@LuigiVampa](https://community.openai.com/u/LuigiVampa)\
**Post date:** [April 18, 2023, 2:47pm UTC](https://community.openai.com/t/do-openai-use-customers-prompts-for-training-their-models/166615/2 "2023-04-18T14:47:10Z")

</div>

I’d hazard a guess and say that the effort required to check the validity of the daft things users input sometimes outweighs their usefulness.  
Not a clue though.

---

<div class="post-metadata">

**Author:** ![shh8362](https://avatars.discourse-cdn.com/v4/letter/s/90db22/32.png) [@shh8362](https://community.openai.com/u/shh8362)\
**Post date:** [April 18, 2023, 2:50pm UTC](https://community.openai.com/t/do-openai-use-customers-prompts-for-training-their-models/166615/3 "2023-04-18T14:50:08Z")

</div>

Makes sense but with the paid plan, I’m not sure if the users still send silly prompts.

---

<div class="post-metadata">

**Author:** ![LuigiVampa](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/luigivampa/32/60316_2.png) [@LuigiVampa](https://community.openai.com/u/LuigiVampa)\
**Post date:** [April 18, 2023, 3:05pm UTC](https://community.openai.com/t/do-openai-use-customers-prompts-for-training-their-models/166615/4 "2023-04-18T15:05:30Z")

</div>

I’ve a paid plan and maybe not silly but certainly plenty of broken ones. Been playing around putting the chat history in as long term memory options and there’s several hallucinated responses and lots of other worthless responses.

So, hope not! Haha

---

<div class="post-metadata">

**Author:** ![shh8362](https://avatars.discourse-cdn.com/v4/letter/s/90db22/32.png) [@shh8362](https://community.openai.com/u/shh8362)\
**Post date:** [April 24, 2023, 10:13pm UTC](https://community.openai.com/t/do-openai-use-customers-prompts-for-training-their-models/166615/5 "2023-04-24T22:13:23Z")

</div>

I found the answer which is available at [https://platform.openai.com/docs/guides/chat/do-you-store-the-data-that-is-passed-into-the-api](https://platform.openai.com/docs/guides/chat/do-you-store-the-data-that-is-passed-into-the-api) under FAQ:

> As of March 1st, 2023, we retain your API data for 30 days but no longer use your data sent via the API to improve our models. Learn more in our [data usage policy](https://openai.com/policies/usage-policies).