Privacy Policy

We record all content you submit and all model outputs (“Data”), including text, images, and files. We also record basic request metadata such as timestamps, model identifiers, and token counts.

We do not store IP addresses, cookies, or account identifiers. We cannot link submissions back to a specific individual.

Data is used to operate the Service, conduct research, evaluate safety, and train models.

Before using Data for model training, we apply best-effort automated redaction to reduce the presence of personal information such as names, contact details, and private keys. However, no automated process is perfect — do not submit anything you would not want potentially included in a training dataset.

We retain Data indefinitely for research and training purposes.

You may request deletion of stored data that has not yet been used for training by emailing [email protected]. Once Data has been incorporated into a trained model, it cannot be individually extracted or deleted. If you do not want your content used for training, do not submit it.