news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

it home reported on september 26 that meta ceo mark zuckerberg said that there are complex copyright issues surrounding the capture of data to train ai models, but he believes that the personal works of most creators are not valuable enough to deserve attention.

in an interview with the verge, zuckerberg said meta might strike “certain partnerships” for valuable content, but if others demanded payment, the company would rather forgo using their content.

"i think individual creators or publishers tend to overestimate the value of their particular content in the overall picture," zuckerberg said in the interview. "my guess is that when the content is really important and valuable, certain partnerships will be established." but if creators feel concerned or object, "when the situation becomes serious, if they ask us not to use their content, then we won't use their content, and it won't have much impact on the outcome of the matter."

meta, like nearly every major ai company, is currently embroiled in litigation over the scraping of data without permission for use in ai training. last year, the company was sued by a group of authors, including sarah silverman, who claimed that meta’s llama model was illegally trained using pirated copies of their work.

meta, like nearly all major ai players, believes that u.s. fair use law should allow this kind of unauthorized scraping. zuckerberg elaborated on the issue:

i think in any new technological medium, there are questions about fair use and the boundaries of control. when you put your work out into the world, to what extent do you still control it, own it, and license it? i think all of these things need to be revisited and re-discussed in the age of ai.

it home noted that microsoft's ai ceo said earlier this year that anything "on the open web" is "free software" and anyone can "copy it, recreate it, and reproduce it."