I don't quite get why this thread was locked, but this would be me unlockin' it.
Because there are no little babies here in need of sheltering and pacification.
In any case, please read the linked article below, which was posted today by Gary Marcus, a noted AI expert/ethicist and Reid Southern, industry artist/scourge of the text to image community. It clearly demonstrates the endemic copyright infringement issues of Midjourney, OpenAI and others - and in the specific case of Midjourney, a purposeful campaign to cover up findings and indemnify the company by changing Terms of Service/shifting liability to subscribers, in real time, as mounting evidence surfaced following the release of its latest version.
It also contains links to a private, leaked text conversation between Midjourney devs and CEO, David Holz, discussing the incorporation of 5,000 new, additional artist names/styles which can be easily accessed by users, as well as discussion of laundering the data to make it untraceable, AND the document itself containing artist names.
I defy anyone to read through all of this in its entirety (complete with output from Chat-GPT that directly plagiarizes entire pages of text from the NY Times) and explain their belief that these models somehow "learn just like humans", or to justify the current business models as ethical, non-infringing, and ultimately in line with
the intent of copyright law.
2024 is the year. A bunch of billion dollar companies are going under, and some execs are headed to prison. End users would be well advised to wise up right now, before doing additional irreparable harm to their careers and reputations.
https://spectrum.ieee.org/midjourney-copyright ...and more converging evidence from TomsHardware.com regarding six different generators, also released today:
https://www.tomshardware...t-copyrighted-characters