...

AI is gathering pace, and in my view, by 2028, it may be too late, so act now.

According to The Atlantic, Meta has made use of datasets that include a sizable number of books that were scraped from resources like LibGen, many of which seem to have been used without permission: https://www.theatlantic.com/technology/archive/2025/03/search-libgen-data-set/682094/

When you factor in the losses of standard licensing fees for publishers, a damages claim has the potential to be very significant. Meta will not want this coming to court so a potential deal can be done which would be life-saving for many publishers. If court action has to follow, then a potential deal can be done, as Meta will not want to incur a large damages judgement and legal fees to boot.

Lawdit is looking for publishers who would like to explore a potential claim in more detail under a potential group litigation/litigation fund with Lawdit and counsel.

The claim would involve significant pre-action work and a number of meetings and discussions with Meta but surely it must be worth exploring?

Lawdit’s view

Meta is going to struggle with any defence it runs in either UK or US law, and in my view most are unlikely to succeed in a UK court due to the commercial nature, extent of copying, and lack of consent involved in the AI training practices. Any data mining or fair dealing exceptions in the CDPA 1988 do not extend to commercial AI training.

Meta is likely to rely on:

  • Fair use analogies (Google Books) Authors Guild v. Google (804 F.3d 202)
  • Implied licence arguments,
  • Technical jurisdictional barriers.

Who knows? But surely we have to try?

Send me an email if you wish to explore this in more detail

michael.coyle@lawdit.co.uk

share this Article

Recent Articles

Written By: