Editorial & AI policy

This page is the standard we hold ourselves to. If an article on this site falls short of it, that’s a bug — tell us.

1. Purpose

Future of Korea exists to answer questions about Korea completely and accurately for readers who don’t read Korean. Every article should leave a reader able to do something: understand a number, make a decision, fill in a form, or ask a better question.

We do not publish content whose primary purpose is to rank in search. If a piece would not be useful to a reader who arrived by accident, it doesn’t run.

2. Sourcing

Primary sources first. Facts come from the institution that produced them:

• Macro data — Bank of Korea (ECOS), Statistics Korea (KOSIS)
• Markets — Korea Exchange (KRX), Financial Supervisory Service (DART), company disclosures
• Trade — Korea Customs Service, KITA
• Immigration and visas — Ministry of Justice / HiKorea, Korea Immigration Service
• Tax — National Tax Service; Health insurance — NHIS; Pensions — NPS
• Policy — the ministry that issued it, in its own release

Dated figures. Every number carries the period it refers to and the date we checked it. “Korea’s fertility rate is 0.75” is useless; “0.75 in 2024, per Statistics Korea’s February 2025 release” is a fact a reader can verify.

Linked. Where a source is publicly accessible, we link it in the Sources section of the article. Korean-language sources are linked and identified as Korean.

No laundering. We do not state a fact on the authority of another outlet’s summary of it. If we can’t reach the underlying source, we attribute explicitly (“as reported by X”) or leave it out.

3. Accuracy and uncertainty

• Where estimates differ, we give the range and say who produced each figure.
• Forecasts are labelled as forecasts, attributed, and dated.
• We distinguish what is announced from what is legislated from what is in force. In Korean policy coverage this distinction does most of the work.
• We do not round in a direction that makes a story better.

4. Use of AI — what is and isn’t allowed

We use AI tools for research assistance, structuring, drafting and language editing. We are open about this because readers are entitled to know how what they’re reading was produced.

AI is permitted to: gather and summarise source material; propose structure; produce a first draft from verified inputs; suggest edits for clarity; generate the abstract artwork on the site.

AI is not permitted to: be the authority for any fact. No figure, date, statute, visa requirement, ticker or quotation is published on the strength of a model’s own recollection. Every such claim must be traced to a named source that our build verifies is reachable, and the key figures in every article are individually cited to those sources so you can check them yourself.

Verification is mandatory; publication is automated. We should be precise about this, because a vague claim would be worse than none. Articles are researched, drafted, fact-checked and published by an automated editorial system, without a person approving each one before it goes live. What stands between a draft and publication is not a human sign-off but a verification gate in our build:

• every statistic in the key takeaways must carry a numbered citation to a named source
• every source must have a publisher, the date it was read, and a URL that resolves when the site is built — a dead link fails the build
• at least one source per article must be a primary one: the statistics office, central bank, exchange, regulator, or the filing itself
• an article that cannot clear those checks is not published

A human reviews published work after the fact, and corrections are issued under our corrections policy. Volume never overrides verification: if the figures cannot be confirmed, the system publishes nothing that day.

No synthetic authority. We do not invent bylines, credentials, headshots, personal anecdotes, or “expert quotes”. Articles are published under the editorial desk because that is literally what wrote them — there is no pretence of a human byline that does not exist.

Imagery. Article images are one of three things, and never anything else:

• charts drawn by our own code from the figures verified in that article, with the data’s publisher named on the image
• licensed photographs from Pexels, credited to the photographer beneath the image. These are illustrative stock photography, not documentary images of the events described
• typographic covers generated by our own code from the article’s headline and lead figure, used when no relevant photograph exists

We do not publish AI-generated photorealistic images of real people, places or events, and we do not present any image as documentary evidence of a news event.

5. Independence

Advertising is sold programmatically through Google AdSense. Advertisers do not see articles before publication, cannot commission or veto coverage, and have no relationship with the editorial desk. Ad units are visually distinct and labelled “Advertisement”.

Where an article discusses a company that may plausibly advertise on the site, that changes nothing about what we write. If we ever accept sponsored or affiliate content, it will be labelled as such at the top of the page, in the same size type as the body text.

6. Corrections

See the corrections policy. In short: we fix errors quickly, in public, with a note saying what changed and when.

7. Updates and freshness

Evergreen explainers — visa rules, tax bands, how-to guides — are reviewed on a schedule and after any rule change. The “Updated” date on an article is real: it changes only when the content changes, never to game a freshness signal.

8. Privacy and reader data

We collect as little as possible and explain what advertising partners collect. See the privacy policy and cookie policy.

9. Complaints

Editorial complaints go to editor@futureofkorea.com with CORRECTION or LEGAL in the subject. We respond in writing.