Big Tech asks Australia to change copyright law to permit machine learning systems to scan copyrighted web sites to construct language models.

Copyright should allow people to study copyrighted works to learn from them, as a matter of basic principle. That includes adducing patterns in how language is used.

If people want a law to prohibit studying works of text or art to make programs that generate text or art, this should be a separate law so that it doesn't mess up other uses of studying works, or other ways of reusing aspects of works. We must not let Big Tech add copyright restrictions, since they will design it to give them more power over us.