How we use AI with your content

• We send AI providers only the text being processed and the minimal context a task needs — never your account, billing, or login data.
• Your content is not used to train any AI model . Requests to OpenAI are sent with storage switched off; DeepL and Google run under no-training business terms.
• You stay in control: delete your account and we erase your content — projects, memory, glossaries, uploaded files and all.

Depending on the language pair and the feature, your text may be processed by one of the following sub-processors. Each is engaged under a written data-processing agreement.

Grammar checking runs on our own infrastructure and does not involve a third-party AI provider.

AI features need context to produce good results, but that context is deliberately narrow. This is the exact boundary.

Sent to the AI provider — only what the task requires:

• The specific segment being translated or scored
• Neighbouring segments, for continuity
• Glossary terms you've configured
• Your brand-voice / tone settings

Never sent — this stays inside Rush Studio:

• Your name, email, or account profile
• Billing or payment details
• Passwords or authentication tokens
• File names, or whole documents in a single request

No. This is the question that matters most, so here is the precise mechanism behind the answer.

OpenAI. Every request is sent with storage explicitly disabled (store: false), which keeps it out of OpenAI's retention for model training and evaluation. Customers on our Scale plan can additionally ask us to arrange an account-level Zero Data Retention agreement, under which OpenAI holds nothing at all once a response is returned.

DeepL and Google Cloud Translation. Both process your text under their standard business / API terms, which contractually exclude using customer content to train their models. Content is processed to return the translation and is not retained for improvement.

Some data is stored inside Rush Studio to make the product work — your working documents, and caches that stop us re-charging you to re-analyse identical text. Nothing AI-generated is kept indefinitely.

Erasure is real, not cosmetic. When an account is deleted — by you from your settings, or by an administrator — we cascade the removal across every store that holds your content.

That includes your projects, translation memory, translation-memory buckets, glossaries, brand voice, comments and annotations, and the original files you uploaded (removed from disk, not just de-listed). The shared analysis caches are not keyed to any individual and so cannot be purged per-person without affecting other customers; instead they expire automatically on the 180-day schedule above.

Under GDPR, the roles split cleanly. Knowing which side of the line you're on tells you what each party must do.

Rush Studio — processor. We process your content only on your instructions, hold DPAs with every sub-processor, apply the retention limits and no-training terms above, and erase your content on account deletion.

You — controller. You decide what content to upload, ensure you have a lawful basis for any personal data in it, handle requests from the people named in that content, and manage who on your team can access a workspace.