Migliaccio & Rathod LLP is investigating whether source code and other content published by users on Replit may have been used to train large language models under terms that users did not clearly understand when they originally created or published their projects.
Replit is a coding platform used by developers to build, host, and share software projects.
Replit’s current terms state that code and content published in public projects may be used by Replit to improve its services, including developing or training large language models. Replit’s commercial terms distinguish public projects from private customer content, which generally is not used to train machine-learning models except where expressly permitted.
What We Are Investigating
The investigation is focused particularly on historical public projects.
Developers may have originally made source code public because they wanted others to view, use, collaborate on, or learn from the project. That does not necessarily mean they understood at the time that their work could later be ingested as training data for Replit’s AI systems.
Migliaccio & Rathod LLP is investigating:
- when Replit first began using public user code for LLM training;
- what Replit’s Terms of Service said when affected projects were originally published;
- whether older terms expressly disclosed AI or machine-learning training;
- whether code uploaded before explicit AI-training language was later incorporated into training datasets;
- how users were notified when Replit expanded its stated uses of public content;
- whether users could prevent older public projects from being used;
- whether deleting or privatizing a project after training began had any effect; and
- whether Replit obtained additional commercial value from developers’ code beyond hosting and sharing the project.
Replit’s current consumer Terms of Service explain that users retain rights in their content while granting Replit certain rights necessary to operate the service. Current terms also contain specific rules governing public and private applications.
Potential Claims May Include:
- unfair or deceptive trade practices;
- breach of contract;
- breach of the implied covenant of good faith and fair dealing;
- unjust enrichment;
- failure to adequately disclose a new commercial use of user-created source code; and
- claims concerning use of software code as AI-training material.
Signs You May Be Affected:
- You used Replit to create or host software projects;
- you published source code or an application publicly;
- your project was published before Replit expressly disclosed LLM training in its terms;
- you did not understand that public code could be used to train Replit AI models;
- your project contained substantial original code or other valuable work;
- you later deleted, privatized, or stopped maintaining the project;
- you used Replit for several years and experienced changes to its Terms of Service; or
- you have older screenshots, emails, archived terms, or project records showing when your code was published.
This investigation is not focused on ordinary private commercial projects where Replit’s current agreements expressly restrict model training. It is particularly focused on whether earlier public-code contributions were repurposed for AI development without sufficiently clear disclosure at the time they were supplied.
If you published original source code through Replit and believe it may have been used for AI training without your informed consent, we would like to hear from you.
Please complete the contact form on this page, send us an email at [email protected], or give us a call.