NepYork treats artificial intelligence as a tool for speed, research, efficiency, and data visualization — never as a substitute for journalistic judgment, original reporting, or public accountability. LLM output is treated with the same skepticism as an unverified source. AI may assist the newsroom; it may not report, decide, or stand behind a story on its own. If an AI-assisted story is wrong, NepYork does not blame the model. The journalist and editor remain fully accountable.
Approved Use Cases
AI may be used in these areas, with human review required before any public use:
1. Research and Investigative Support
- Semantic search across large document sets
- Summarizing lengthy records to surface leads
- Identifying patterns in structured data
- Early-stage research assistance
AI findings are leads only. They are not facts until independently verified.
2. Administrative, Editing and Production Tasks
AI may assist with routine editorial and production tasks, including:
- Correcting grammar, spelling, punctuation, and basic syntax
- Improving clarity and readability without changing the substance of a story
- Audio and video transcription
- Translation drafts
- Shot lists, meeting alerts, and internal notes
- File organization and workflow support
AI-assisted grammar and copy edits must be reviewed by a NepYork journalist. AI may correct language and improve readability, but it may not introduce, remove, or alter facts, quotes, context, attribution, or the meaning of a reported story.
3. Packaging After Reporting Is Complete
Editors may use generative AI to draft:
- Headline options
- Social media copy
- Newsletter summaries
- SEO-oriented variants
The published version must be selected, rewritten, and approved by a human editor.
4. Brainstorming
Journalists may use AI to:
- Overcome writer’s block
- Generate interview-question angles
- Outline possible story structures
5. Data Visualization and Labeled AI Imagery
NepYork may use AI to visualize actual, verified data — for example, through charts, maps, diagrams, explainer graphics, or simplified illustrations derived from reported numbers and documents.
NepYork may also generate images for illustration, explainers, or conceptual accompaniment only if:
- The image is clearly labeled as AI-generated.
- The image does not pretend to be a news photograph, eyewitness image, or documentary record.
- The underlying facts being illustrated have been verified by journalists.
- An editor approves the image before publication.
Acceptable label examples include:
- AI-generated image
- AI-generated illustration based on verified data
- AI visualization of reported figures
The label must be visible to the reader in the caption, credit line, or immediately beside the image.
Prohibited and Restricted Practices
| Practice | NepYork rule |
|---|---|
| Writing core news stories | Prohibited. Published articles must be reported and written by journalists. |
| Presenting AI images as real news photos or video | Strictly prohibited. AI may not create, alter, or enhance documentary photography or video, or pass off generated visuals as captured reality. |
| Using AI as a source | Prohibited. AI output cannot serve as a source or verified fact on its own. |
| Unlabeled AI images | Prohibited. Any generated image or AI data visualization must be labeled. |
| Visualizing unverified data | Prohibited. AI graphics may represent only data that journalists have checked. |
| Automated publication without review | Prohibited. |
| Changing facts, quotes, attribution or meaning through AI editing | Prohibited. AI may correct grammar and improve readability but may not alter the substance of reported journalism. |
Governing Principles
Every word, headline, caption, graphic, data point, or image suggested or modified by AI must be checked and approved by a NepYork journalist before publication.
Errors in AI-assisted work are newsroom errors. Responsibility rests with the reporter and editor.
Staff should treat LLM output as they would an unconfirmed tip from a stranger.
AI visuals must serve reported facts. They may not invent events, faces of real people as if photographed, crime scenes, protest scenes, or “what it looked like” reconstructions unless clearly labeled as illustrations and editorially justified.
Disclosure
When AI plays a material role in published content — including generated images or AI data visualizations — NepYork will disclose that fact in plain language.
Rules
- Do not paste confidential sources, unpublished documents, or unpublished reporting into public AI tools unless an approved internal system is used.
- Do not ask AI to invent quotes, scenes, statistics, or unnamed sources.
- Do not use AI to impersonate a source, official, or member of the public.
- Do not use unlabeled AI-generated images, voices, or video.
- Do not use AI to alter real news photographs or footage.
- AI images of real, identifiable people are restricted and generally avoided unless there is a clear editorial reason, the image is not presented as a photograph, and it is labeled.
- AI may be used to correct grammar, spelling, punctuation, and basic syntax, but all such changes must be reviewed by a journalist and must not alter facts, quotes, attribution, context, or the meaning of the story.
- If an AI visualization misstates the data and is published, the correction will be handled as a standard journalistic correction.
NepYork may use AI to visualize verified data and to generate labeled illustrations. It may not use AI to fake reality. NepYork may use AI to work faster and write more clearly. It may not use AI to decide what is true.