\n\n
The most recent developments in AI copyright licensing focus on three fronts: publisher coalitions, court rulings, and national legislation.
- Publisher coalition growth – More than 80 publishers in over 20 countries, including The New York Times, AFP and The Economist, have banded together in the Standards for Publisher Usage Rights (SPUR) coalition. SPUR is rolling out a Content Telemetry Standard that will let AI firms log exactly how much news content they consume, making usage‑based licensing feasible and giving publishers a clearer path to compensation.
- Court decisions shaping fair‑use limits – In the U.S. Third Circuit, the court reaffirmed that copyright‑protected legal headnotes cannot be used to train competing AI without a license, emphasizing that ordinary fair‑use analysis still applies to AI training data. This underscores that AI developers must consider existing licensing markets when sourcing copyrighted material.
- Legislative moves – Two jurisdictions have taken opposite approaches. India’s Delhi High Court recently refused an interim injunction against OpenAI, saying that merely storing news content for training does not automatically equal infringement. By contrast, Saudi Arabia’s 2026 copyright law introduced Article 26, an explicit exception that allows AI developers to train on any lawfully obtained published work without permission or payment, creating a statutory safe harbour for AI training.
Why it matters – Together these trends signal a shift from ad‑hoc lawsuits to industry‑wide frameworks. Publishers are moving toward standardized, measurable licensing, courts are clarifying that AI training is not a blanket exception to copyright, and governments are either tightening or loosening the rules, which will affect where AI companies source data and how they negotiate rights.
What to watch next –
1. Adoption of SPUR’s telemetry standard by major AI providers and the emergence of licensing deals under that model.
2. Follow‑up rulings in the U.S. that may address generative‑AI models (as opposed to narrow legal‑research tools) and the scope of fair use.
3. Implementation of Saudi Arabia’s Article 26 regulations and early enforcement actions, which could set a precedent for other nations considering explicit AI‑training exceptions.
4. The outcome of the ANI v. OpenAI case in India, which may provide judicial guidance on the limits of copyright for AI training data in common law jurisdictions. [[1, 2, 3, 4]]