I wrote about a big new study showing that as much as 25% of high-quality data from popular AI training sets has been restricted in the past year, as publishers scramble to update robots.txt and change their terms of service to prevent AI scraping.
https://t.co/6v0FGQwDYc