JOURNALISM VERTICAL

Journalism sits at the centre of the AI-training debate — news archives are prime training corpora, and AI systems summarise, regurgitate and sometimes mis-attribute reporting. News archives are among the most contested training corpora, which puts the training-ingestion and text-and-data-mining questions front and centre. Beyond training, AI summarisation and snippet generation raise distinct questions about substitution and attribution — a summary that displaces the original is different from one that points to it. Journalism also carries concerns most other verticals do not: attribution and byline (an author’s moral interest in being credited), source protection, syndication across many outlets, and a rising volume of AI-assisted or AI-written articles whose authorship and accuracy need to be legible.

The training-ingestion / TDM opt-out fields are the core here — they let a publisher declare its position on AI training. Output-licence can distinguish search indexing from AI input. Source-material records derivation, output-provenance flags AI-written content, and the CDR carries lineage. What is missing is structure for attribution, summarisation and syndication.

1 — Attribution / byline declaration (QUEUED): a machine-readable record of required credit and author moral interests. 2 — Summarisation & snippet permissions (QUEUED): the position on AI summaries and extracts, separate from training. 3 — Source-protection handling (QUEUED): how confidential-source material is marked so it is not exposed downstream. 4 — Syndication scope (QUEUED): declaring permitted redistribution across outlets and aggregators.

Legal references on this page are to well-established general doctrines only — no specific statutes, section numbers, dates, or figures are asserted. Qualified counsel should review before any of this is treated as authoritative. The field families listed above are not yet specified.

Source Material Architecture (v3.34) for derivation and third-party content. Output-Provenance Architecture (v3.17) for AI-written flagging. TDM opt-out status via CIP-TDM-Opt-Out. cip.md Generator for operator declarations.