Totally understand the concerns about copyright protection. However, I think it's important that UK aligns its copyright and AI laws with those of our European neighbours. The opt-out solution is not perfect, and currently it's technically challenging. However, the Internet Task Force is working on a robots.txt based solution which should be implemented.
The tricky thing for the UK government is they've said they won't go ahead unless and until there's reliable technology in place for creators to manage opt-out arrangements.
I suppose that it depends on how “reliable” the technology is. The robots.txt technology essentially depends on the voluntary commitment of the developers of crawler engines not to index the websites, or at least not to make publicly available the results.
I think that the same will happen with AI training data, except it is going to be much more difficult to demonstrate that copyrighted content was illegally used in training, particularly as the US companies are arguing that this training is “fair use”.
On the other hand, it seems to me that the “threat” of introducing a TDM exemption may lead to copyright owners setting up licensing deals, as has happened in Europe (e.g., Springer-Verlag).
Totally understand the concerns about copyright protection. However, I think it's important that UK aligns its copyright and AI laws with those of our European neighbours. The opt-out solution is not perfect, and currently it's technically challenging. However, the Internet Task Force is working on a robots.txt based solution which should be implemented.
The tricky thing for the UK government is they've said they won't go ahead unless and until there's reliable technology in place for creators to manage opt-out arrangements.
I suppose that it depends on how “reliable” the technology is. The robots.txt technology essentially depends on the voluntary commitment of the developers of crawler engines not to index the websites, or at least not to make publicly available the results.
I think that the same will happen with AI training data, except it is going to be much more difficult to demonstrate that copyrighted content was illegally used in training, particularly as the US companies are arguing that this training is “fair use”.
On the other hand, it seems to me that the “threat” of introducing a TDM exemption may lead to copyright owners setting up licensing deals, as has happened in Europe (e.g., Springer-Verlag).