How we use AI and handle your content
Does our AI translation train on your data? No. Your content is processed only through business AI services whose terms bar training on it.
You stay in control of where AI is used, content type by content type. This page sets out where AI is used in our work, where people decide, and how we handle the content you send, from a free check to a monthly program.
Where AI is used and where people decide
- AI agents translate against your glossary, translation memory (TM), and style rules, then run defined checks before and after the linguist.
- A native linguist works on every job, at the depth your service level sets. At AI + Check, a native linguist checks every flagged segment plus a random sample. At AI + Edit, a native post-editor edits every segment. At AI + Edit + Review, a second linguist also reviews every segment and reads the delivered file.
- A linguist decides every finding. At AI + Edit + Review, a judge agent recommends first. Your instructions outrank both.
- We use in-house AI tools built by Locstars on leading AI models.
- Your quote names the service level for each content type, so you know where AI is used before work starts.
Our responsible AI use statement
We use AI where it does the volume work well, and we keep the decisions with people. These principles apply to every job on this site.
- People decide. Agents propose and check. A linguist decides each finding, and your instructions come first.
- Evidence, not opinion. A finding counts only when it names a concrete defect and points to a reference: the source, your glossary, a rule, or your instructions.
- Depth follows risk. Regulated and high-risk content runs at AI + Edit + Review or without AI, never at AI + Check.
- No delivery without a linguist. Agents do not deliver or publish a file on their own.
- No shortcuts at the gate. If the quality gate flags something that is not an error, we fix the gate’s rule and record the case in your delivery note. We never edit your file just to make it pass.
- Your rules stay yours. Each client’s rules live in that client’s rulebook only.
- Your choice of no AI. You can ask for any content type in your program to be handled without AI.
How your content is processed
Your content is processed only through business AI services whose terms bar training on it. This applies to text you send for a free check, files you send with a quote request, and every job after that.
We work from your exports, or inside your translation management system (TMS), including Phrase, XTM, Smartling, Crowdin, and Lokalise.
For a free check or a quote, please remove personal data from your text and files.
Content that contains personal data
Support macros, help articles, and user reviews can contain personal data, such as names, email addresses, or order details. Tell us if your content contains personal data; we handle it under your instructions. You can also choose the no-AI route for that content. We sign your data processing agreement (DPA), as we sign your NDA.
Choosing no AI
Prefer no AI? Tell us, and your content is translated by native linguists without machine translation or generative AI, using your translation memory and glossary. You can choose it for a whole program or for specific content types, and the quote names it. The no-AI option is available in every language we list.
Confidentiality
We sign your NDA before work starts. Send it with your quote request, or ask us to sign it before you share any files.
Your translation memory
You own your translation memory. We hand it over as a Translation Memory eXchange (TMX) file whenever you ask.
Deleting your files
We delete your files on request. Email info@locstars.com and tell us which project, free check, or quote request the files belong to.
Questions about this page
If your security or privacy review needs answers this page doesn’t give, send your questions or your questionnaire to info@locstars.com. Under your NDA, we tell you which AI services process your content.