Theft is not fair use

Theft vs Copying Terminology

• Large subthread disputes calling copyright infringement “theft.”
• Many argue: copying doesn’t deprive the owner of the original, so it’s not theft but infringement; “theft” is emotional rhetoric used by rightsholders.
• Others counter: what’s “stolen” is income, opportunity, or ownership claim over IP; morally it feels like theft even if legally distinct.
• “Identity theft” is criticized as misdirection from authentication failures; language can shift blame to victims.

Piracy, Harm, and Artists’ Livelihoods

• One side: piracy and AI training undermine creators’ ability to get paid, pushing art toward being a hobby for the wealthy.
• The opposing view: no one has a right to income in a specific field; open source and “sharing” show alternatives.
• There’s disagreement whether saying “piracy isn’t theft” is nitpicking or essential to keep moral categories clear.

AI Training, Fair Use, and the Law

• Dispute over whether training on copyrighted data without consent is fair use or mass infringement.
• Some argue law targets distribution of copies, not “using” them to learn; training is analogous to research.
• Others cite “harm to the market” and “amount taken” tests (e.g., UK fair dealing) and claim commercial AI training clearly fails them.
• A recurring theme: if OpenAI’s scraping is illegal, it should be fined; if not, piracy is effectively decriminalized, exposing a corporatocracy.

Human Learning vs AI Learning

• Pro‑AI commenters analogize models to humans reading books, learning styles, then creating new works; if that’s legal for a person, why not for a machine?
• Critics respond that scale and commercialization matter: one trained model serves millions, unlike individually trained humans, and the product wouldn’t exist without others’ works.

Scale, Power, and Corporate Control

• Concern that tech giants “scalp” culture, build monopolistic moats, and starve original sources (e.g., news sites) by summarizing content.
• Some see two legal regimes emerging: permissive for corporations, strict for individuals.

Cultural Impact of AI Models

• Worry that models encode median, mass‑market culture, weakening incentive and visibility for fringe/experimental art.
• Others think fringe work will still exist for “committed freaks” and can be surfaced with the right prompts.

Debate over the Example Image

• Several question the article’s image evidence: without showing prompts, similarity may just reflect requested composition/styles.
• Many argue the AI output is not a literal copy; overall composition similarity alone is likely non‑infringing.