The Register says the more we learn about how AI does business, the more unfair its practices look.
The Register’s column criticizes big AI’s approach to content, framing the problem as “take the work, keep the money.” It argues that closer scrutiny makes the business look increasingly unfair.
861
1 Source, first seen 2h ago
#17
#17
Not enough discussion yet.
No sentiment analysis available yet.
The dispute over how generative AI companies obtain and use creative work is moving through courts, contracts and the economics of publishing. A column in The Register argues that the central imbalance is straightforward: AI systems depend on books, journalism, images, code and websites, while creators often receive neither payment nor usable information about how their work was used.
The column points to recently unsealed material in The New York Times' lawsuit against OpenAI and Microsoft. The cited statements appear in the news plaintiffs' summary-judgment arguments, and the court has not ruled on those claims. OpenAI and Microsoft dispute the allegations and argue that training on public material is fair use that advances knowledge rather than creating an unlawful market substitute.
That distinction matters. Internal comments quoted by the plaintiffs may illuminate how employees thought about training data, but they are evidence being contested in litigation, not a final judicial finding.
The Register also examines a recent Ninth Circuit decision in the Doe v. GitHub case. The court held that creating new code without copyright-management information is not necessarily the same act as removing or altering that information from an existing work. The ruling did not decide that every use of open-source code for model training is lawful, that generated code cannot infringe copyright or that open-source licenses no longer apply.
For developers and companies using generated code, the unresolved problem is provenance. A suggestion may arrive without identifying the source project or license terms that shaped it. For writers, publishers and artists, the parallel problem is compensation and substitution: systems trained on their work can produce answers that reduce the need to visit or pay the original source.
The column's conclusion is an argument, not a settled legal result. Courts are still deciding where fair use, licensing and market harm meet, while AI companies and rights holders continue to negotiate outside the courtroom.
Not enough discussion yet.
No sentiment analysis available yet.
