AI‑art debates are loud online
A viral thread arguing for “no AI slop” in art production drew about 10,510 likes, while another post critiquing historical photobashing picked up 5,199 likes—signals that the community is actively arguing about technique, ethics and authorship right now (x.com). Those engagement numbers show this debate isn’t niche: artists, galleries and buyers are watching social sentiment as they decide whether to embrace or reject AI processes (x.com).
The fight over artificial intelligence art is no longer just artists yelling at each other on social media. In February 2025, Christie’s launched “Augmented Intelligence,” its first auction devoted to artificial intelligence art, and more than 3,000 people signed an open letter asking the auction house to cancel it before bidding opened on February 20. (theartnewspaper.com) The complaint was not that computers had entered art for the first time. The complaint was that some image models were trained on copyrighted work without permission, and critics said Christie’s was turning that pipeline into a luxury product with estimates above $600,000 for the sale. (theartnewspaper.com) (forbes.com) That is why online arguments now split into two different questions. One question is whether a finished image looks lazy or derivative, and the other is whether the system behind it was built from other people’s labor in the first place. (forbes.com 1) (forbes.com 2) United States copyright policy has moved in a way that sharpens that split. On January 29, 2025, the United States Copyright Office said generative artificial intelligence outputs can be copyrighted only when a human author determines enough of the expressive elements, and that prompts alone are not enough. (copyright.gov) (newsroom.loc.gov) That rule makes authorship sound less like “who typed the request” and more like “who actually arranged the picture.” If a person paints over, edits, sequences, or otherwise shapes the result in a visible way, the Copyright Office says that human contribution can still count. (newsroom.loc.gov) The training fight is moving on a different track in court. In June 2025, a federal judge ruled that Anthropic’s training on lawfully acquired books was fair use, while still sending questions about a separate library of roughly 7 million pirated books to trial. (cnbc.com) Visual art has its own cases, and they are still unresolved enough to keep everyone on edge. The Andersen v. Stability AI lawsuit filed by artists in January 2023 was still active as of April 6, 2026, and Getty Images’ United Kingdom case against Stability AI produced only limited clarity in a November 4, 2025 judgment, including a trademark win for Getty rather than a broad answer on training legality. (courtlistener.com) (newsroom.gettyimages.com) (paulweiss.com) Companies have started building product rules around that uncertainty instead of waiting for a final legal answer. OpenAI says DALL·E 3 declines requests for images “in the style of a living artist,” and it also offers creators a way to opt images out from training of future image models. (openai.com) Meanwhile the audience has become better at spotting what it dislikes. Forbes reported on April 9, 2026 that convention-goers, comics readers, game players, and ad viewers are now reacting not just to obvious mistakes in artificial intelligence images but to the mere suspicion that a supposedly human work may have been traced from one. (forbes.com) That is why a loud post about “artificial intelligence slop” can travel so far now. The argument is carrying three separate fears at once: that the image is cheap, that the credit is blurry, and that the training data may include unpaid work from the very artists being pushed out of the market. (forbes.com) (copyright.gov) (theartnewspaper.com)
