@nishio: I had a sense of dĂ©jĂ  vu when someone started saying “no AI learning”
 I was thinking, “no unauthorized links”.

@fladdict: a well put together article by a real professional. Long but important for everyone to read. Reading
 Midjourney, Stable Diffusion, mimic and other image auto-generating AI and copyright | STORIA Law Firm

  • @nishio:

  • “I will prohibit you from using my illustrations for AI learning”
 In the first place, such a statement may be invalid in violation of Article 30-4 of the Copyright Act, and even if it does not violate Article 30-4 of the Copyright Act, a unilateral statement cannot be concluded as a contract. such a representation is meaningless, at least from a legal standpoint.

@hrjn: if you combine crawler avoidance and warning people who try to download manually, I think it could be prevented in a different way than copyright. It’s illegal to circumvent technical protection measures. However, I’m not sure to what extent this applies, since I think the usual means of protection refers to encryption and CRM.

  • @nishio: @hrjn I’d like to leave the detailed discussion to the experts

  • In the case of provisions restricting rights for public interest purposes, it is considered that the public interest takes precedence over the will of the copyright holder, etc. Therefore, even if the use involves circumvention of technological protection measures, it cannot be said at present that the economic interests of the copyright holder, etc. are likely to be significantly impaired.

  • (7th Meeting in 2008) Minutes | Agency for Cultural Affairs

  • I feel that it is a borderline where I can’t participate in the discussion with my miscellaneous knowledge, because this time “circumvention of technical protection measures is illegal” or “you can override if you sign a contract” which I thought before reading the original article is denied by the experts.

  • The denial of “You can override if you sign a contract, right?” is the part here.

In the discussion of fine tuning, I saw the claim that fine tuning is not information analysis because it does not use numerous works. After checking, we found that “information analysis” is defined in the second paragraph of Article 30-4 as “extracting, comparing, classifying, or otherwise analyzing information pertaining to language, sound, images, and other elements that constitute said information from a large number of works or other large amounts of information. So, the argument is that fine tuning with a few images of a single author is not “information analysis” because it does not satisfy the “many” requirement. I certainly agree~.

This page is auto-translated from /nishio/AIć­Šçż’ăŠæ–­ă‚Š using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I’m very happy to spread my thought to non-Japanese readers.