Freesound's per-uploader AI training preference is a transparency signal, not a legal restriction

In February 2026, Freesound added an account-level setting for uploaders to express preferences about whether their sounds may be used for AI model training, with options ranging from ‘follow CC recommendations’ to ‘open-source only’ to ‘no commercial models’. A common misreading is that these preferences override or supplement the CC license. The FAQ explicitly states they are ‘not legally binding and are not part of the Creative Commons license’. They are voluntary transparency signals intended to help model developers understand uploader intent. The legal obligation comes solely from the CC license. Model developers who ignore a stated preference are not technically violating the law — but Freesound frames compliance as an ethical expectation.

Examples

An uploader sets their AI preference to ‘no commercial models’ but their sounds are under CC0. Legally, a commercial AI lab could still train on those sounds. Ethically, they are acting against the uploader’s stated wish.

Assessment

True or false: if an uploader sets their Freesound AI preference to ‘open-source only’, a commercial company is legally prohibited from using those CC0 sounds to train a proprietary model. Explain your answer.