AI & Sourcing
Policy

Our AI Philosophy

Artificial intelligence is transforming journalism. We embrace this change — selectively, transparently, and with an unwavering insistence that human judgement, human accountability, and human editorial standards govern every word published under THE CELEBRITY masthead.

We do not use AI to replace journalists, generate articles autonomously, fabricate sources, or simulate interviews. We do use AI to help our journalists work faster and smarter — processing large datasets, surfacing relevant archival material, supporting translation workflows, and drafting first passes that are then substantively rewritten by human writers.

This policy applies to all full-time editorial staff, contributing writers, freelancers, and Brand Studio content creators working on editorial-adjacent projects. Violation of this policy constitutes a serious disciplinary matter.

Approved AI Uses (with Required Disclosure)

The following uses of AI are approved within our editorial workflow, subject to the disclosure requirements detailed in Section 7.

Research & Discovery

• Searching and summarising large volumes of Korean-language press releases, agency statements, and regulatory filings.
• Identifying relevant archival articles within our content management system.
• Generating structured research outlines for long-form journalism.
• Analysing chart data, streaming figures, and commercial performance metrics at scale.

Drafting Support (Subject to Substantial Human Rewrite)

• Producing first-draft structures for news articles based on verified facts provided by the journalist.
• Suggesting headline and deck variations for human editorial selection.
• Generating initial SEO metadata drafts, subject to human review.

Translation Assistance

• First-pass Korean→English translation of source materials, to be reviewed and edited by a Korean-fluent staff member or contractor. (See Section 8 for full translation standards.)

Data Journalism

• Processing publicly available datasets (streaming figures, award votes, chart histories) to identify patterns and trends, with human analytical interpretation.

Prohibited AI Uses

• Publishing AI-generated articles without substantial human rewrite — no article may be published in which the AI-generated content constitutes more than 10% of the final word count.
• Fabricating or simulating quotes — AI must never be used to generate, paraphrase, or simulate statements attributable to real individuals.
• AI-generated interview responses — we do not publish "AI-imagined" interviews or "what they might have said" content.
• Bypassing the correction process — AI tools may not be used to silently re-edit published articles without editorial review and correction disclosure.
• Generating false images of real people — deepfake imagery, AI-generated likenesses, and synthetic celebrity images are never used in editorial content.
• Using AI to identify or aggregate private information — AI may not be used to scrape, aggregate, or infer personal data about private individuals or non-public aspects of public figures' lives.
• Presenting AI analysis as verified fact — AI-generated analysis, predictions, or trend interpretations must be clearly framed as analytical inference, not confirmed reporting.

AI Use-Case Matrix

The following matrix summarises how AI may be used across different content types. Green = approved with disclosure; Amber = conditional on senior approval; Red = prohibited.

Conditional uses require documentation of the approving senior editor in the article's content management record.

Approved AI Tools (March 2026)

The following tools have been evaluated and approved by our Technology & Editorial Review Committee. Staff must use only approved tools for editorial work. Personal AI subscriptions used for professional content must be disclosed.

AI-Assisted Editorial Workflow

The following flow illustrates the required human checkpoints when AI tools are used in article production. AI may assist at stages 1 and 3; human editorial control is mandatory at all other stages.

Disclosure Standards

When AI tools have contributed meaningfully to an article's research, structure, or translation, readers are entitled to know. THE CELEBRITY's disclosure standards are as follows:

Standard AI Disclosure Label

When Disclosure Is Required

• AI was used to generate or substantially structure any part of the published text.
• Machine translation was used on a primary source that is directly quoted or paraphrased.
• AI-generated artwork or illustration is used (must include "AI-generated illustration" caption).
• Data analysis was AI-assisted and forms a material part of the story's argument.

When Disclosure Is Not Required

• AI was used only for background research that did not find its way into the article.
• Spellcheck or grammar tools (Grammarly, etc.) were used for copy editing.
• AI-powered content management or distribution tools (not content creation).

Machine Translation Standards

Korean-to-English translation is one of the most important and sensitive AI applications in our workflow. Mistranslations in entertainment journalism — particularly involving artist statements, agency communications, and cultural nuance — can cause significant harm.

Required Translation Protocol

1. Machine first pass: Approved translation tool (DeepL Pro preferred) produces initial translation.
2. Journalist review: The reporting journalist reviews the translation against context and flags uncertain terms.
3. Korean-fluent verification: A Korean-fluent staff member or vetted contractor reviews any direct quotes, culturally specific terminology, or statements that carry legal or reputational significance.
4. Source notation: Translated material is tagged in the article metadata with "source: Korean-language, machine translation verified by [staff member]."

Cultural & Linguistic Edge Cases

Korean honorific language (존댓말/반말), fan community terminology (stan culture vocabulary), company-specific idol training terminology, and idiomatic expressions in idol fan café posts require particular care. Machine translation of these contexts is treated as unreliable by default and must be verified by a specialist.

Our Seoul bureau team provides daily terminology guidance and is available for urgent translation escalations at seoul@thecelebrity.com.

AI-Generated Images & Visual Media

The use of AI image generation in editorial contexts raises significant ethical, legal, and representational issues. Our policy is deliberately conservative.

Editorial Content

• Photographs of real artists must be from licensed sources (Getty Images, our contracted photographers, or agency-provided press materials). AI-generated images of real artists are never used in editorial contexts.
• Abstract conceptual illustrations (e.g., chart visualisations, cultural concept art, event backgrounds) may use AI generation with explicit "AI-generated illustration" captioning and Senior Editor approval.
• AI image tools may not be used to alter, de-age, re-age, or otherwise manipulate licensed photographs of real individuals.

Brand Studio Content

• AI-generated imagery in branded content requires written approval from the partner brand, disclosure in the content, and Head of Brand Studio sign-off.
• AI-generated likenesses of any real individual, including brand ambassadors and K-pop artists, are prohibited in commercial content regardless of client request.

Accountability Framework

Every AI-related editorial decision has a human owner. The following accountability structure ensures that AI use is monitored, audited, and continuously improved.

Policy Version History

This policy is reviewed quarterly by the Technology & Editorial Review Committee and updated as AI capabilities and industry standards evolve.

Questions, concerns, or suggestions for this policy should be directed to ai-policy@thecelebrity.com. We welcome feedback from our readers, journalists, and industry peers.