ConstraintCognitive automation2025-07-01
Crunchyroll's German subtitles for an anime premiere contained the line 'ChatGPT said…'; the company said a third-party vendor had used AI-generated subtitles in violation of its agreement
Translator / Interpreteroccupation page →
Event date / reported
2025-07-01 · reported 2025-07-03
Evidence stage
ConstraintFailure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.
Tasks this bears on
First-draft translation of routine text
Manuals, product copy, correspondence, internal documents.
Automating✓ Evidence-backed
Post-editing machine output
Fixing what the machine got wrong, and knowing where to look for it.
New task✓ Evidence-backed
Where this applies
Entertainment subtitling, one streaming platform, one episode (Necronomico and the Cosmic Horror Show, ep. 1, 1 July 2025). The company's president had said in 2024 it was testing generative AI for subtitling; the failure was unreviewed vendor output.
What this means
A constraint record that touches both linked tasks: a streaming platform's German subtitles shipped with 'ChatGPT said…' in them, and the platform's response was that its vendor had used AI subtitles in breach of contract. On the deployment layer it shows a buyer who contractually forbids raw machine output, and it shows the cost of skipping post-editing in public — the first draft is cheap, the unreviewed first draft is not.
What it does not yet show
One episode on one platform, in entertainment subtitling — tight timing, vocal fans — does not describe technical, legal or corporate translation. The same company had said it was testing generative AI for subtitles, so the record is about who may use it and who checks it, not whether MT is used; and a contract clause is not evidence that post-editing pays translation rates.
What you can check
Read the AI/MT clause in your current agency or client contract (or ask for it if you have never seen one) and note three things: whether machine output is allowed, whether you must disclose using it, and what the post-editing rate is relative to translation; those three lines are this record applied to your own work.
Does it change the assessment?
No. The impact index is never moved by a single event. What this record did: the 2 linked task judgements above now rest on evidence instead of inference.
Source
Engadget · verified 2026-09-10 · Claude (VOLO agent) — source text fetched and cross-checked · interpreted 2026-09-10 · Claude (VOLO agent)