> One possible best practice would be to allow authors to specifically opt out of use of their output for ML training. If it’s a copyright matter, I don’t see

> One possible best practice would be to allow authors to specifically opt out of use of their output for ML training. If it’s a copyright matter, I don’t see how that could work. It’d need to be opt-in, or covered by an explicit license grant (and terms and conditions are being ignored to the point that I gather some jurisdictions’ courts are pretty much striking down anything that a reasonable person wouldn’t expect to be there, and so it wouldn’t surprise me if such an approach would strike down any grant asserted in T&Cs). > for software authors, prohibiting ML training would be antithetical to the Open Source Definition If it’s a copyright matter, this wouldn’t be the case, because it wouldn’t be discrimination against a particular field of endeavour, but rather simply insisting on the terms of the license.