Data types
How do I de-identify source code for AI training?
By SourceX Editorial · Updated
Short answer
To de-identify source code, remove or replace secrets and API keys, internal hostnames, author emails and third-party or customer-owned code, check free text and attachments for hidden details, review samples by hand and exclude anything doubtful. Your company then approves the prepared copy before it leaves.
What source code usually contain#
Typically repositories, commits and history. The work itself is what buyers value; the identities are not.
What to remove or replace#
At minimum: secrets and API keys, internal hostnames, author emails and third-party or customer-owned code. Names are often swapped for consistent placeholders so conversations still make sense.
How to check it worked#
Automated tools catch most details, but not all. Reviewers spot-check samples, look for rare facts that could point to a person, and drop records they are unsure about. Buyers are also contractually barred from trying to identify anyone.
How SourceX handles it#
SourceX starts with a short fit check that shares no data. If it fits, rights are reviewed, a copy is prepared with personal and confidential details removed, and your company approves exactly what leaves before anything is delivered under a signed license.
Frequently asked questions
Can I do the de-identification myself?
You can, but SourceX or an approved preparation partner normally does it and you review the result.
Is this legal advice?
No. It is general information; your counsel should review any specific deal.
Related resources
- QuestionDo AI labs buy code?
- InsightHow do I de-identify CAD and engineering drawings for AI training?
- InsightHow do I de-identify code reviews and pull requests for AI training?
- IndustryBPO & contact centers data
- IndustryRecruiting & staffing data
- DataCode review records
- DataSource code
- QuestionShould companies sell or license their data?
- InsightIs it safe to license company data for AI training?
- SolutionData partnerships between businesses and AI developers
- SolutionWhat is AI evaluation data?
- GlossaryAI training data
See if your company qualifies
A short company assessment. No data uploads are needed.