We don't train AI on your code.
It is our standing commitment, and the architecture backs it up: there is no shared corpus to train on, and you can leave any time.
"We do not use your repositories or CI jobs to train or improve machine-learning models, and we do not access their contents except as necessary to provide, secure, and support your instance."
No shared corpus
Your instance is yours alone, not a shared tenant beside other customers' code. There is no cross-customer corpus to train on in the first place. The strongest assurance is structural, not a promise.
The exit is the guarantee
Forgejo is open source and your data exports any time, in the upstream format. If we ever broke this, you would not be trapped here. That is the point.
Honest about third parties
We run on third-party infrastructure and use analytics, which see operational data, not your repositories, under their own terms. Your code isn't sent anywhere to train a model.
In plain words
We do not use your repositories or CI jobs to train or improve any machine-learning model, and we do not read their contents except as necessary to run, secure, and support the service for you. Your code is yours. We host it, we do not mine it. And because your instance is single-tenant and you can export and leave at any time, you don't have to take that purely on faith.