Training has not, in fact, been decided on yet. Of the four criteria deciding whether something is fair use, training only maybe passes two.
https://www.copyright.gov/ai/Copyright-and-Artificial-Intell...
https://www.copyright.gov/ai/Copyright-and-Artificial-Intell...
This order grants summary judgment for Anthropic that the training use was a fair use.
And, it grants that the print-to-digital format change was a fair use for a different reason. But it
denies summary judgment for Anthropic that the pirated library copies must be treated as
training copies.
Various uses of copyrighted works in AI training are likely to be transformative. The
extent to which they are fair, however, will depend on what works were used, from what
source, for what purpose, and with what controls on the outputs—all of which can affect the
market.
That is not what the text you quoted is saying. It says that the output may be transformative, which is one criteria, but the other criteria depends on how it is used and what the source is.