Artificial Intelligence Policy

I. Core Principle

Artificial intelligence tools help the editorial team at «NikVesti» to work more efficiently, but decisions regarding the content of publications are made by people. The author and editor are responsible for every published piece, regardless of which tools were used during the production process. These rules apply to all editorial staff, as well as to freelancers and external contributors.

Responses from chatbots and other AI systems are not considered sources of information. Facts, figures, quotes and comments obtained using AI are verified against the original source before publication.

II. Where the editorial team uses AI

Machine translation. Russian and English versions of articles are generated using the DeepL machine translation service and published immediately. A translator and proofreader checks them after publication. Until the translation has been verified by a human, a note to this effect appears beneath the article’s headline.

SEO headlines and SEO descriptions. An editor can generate these using a language model within the content management system, after which they check and edit them as necessary.

Editorial recommendations for drafts. As part of a pilot project by International Media Support (IMS), journalists can submit a draft for review by an AI tool, which returns up to three recommendations regarding structure, clarity and compliance with our standards. The tool does not publish anything: the author and editor jointly decide whether to accept or reject the recommendations.

Proofreading. Published articles are automatically checked for spelling mistakes using dictionaries; the language model merely confirms or rejects the suggested corrections. The author is notified of any errors and makes the corrections themselves.

In-house AI editorial assistant. It monitors government documents, public procurement, and law enforcement announcements, and assists with analysis and searches in our archive. Everything it finds serves as a lead for journalists.

Social media. AI helps prepare drafts of posts and slides for carousels based on our published content. Every post is checked and edited by a designated member of the social media team before publication.

Journalists’ tools. Journalists can use audio transcription services and language models to support their work. Everything created with their help is checked by a human before publication.

III. What we do not do

We do not publish AI-generated texts without editorial review.

We do not use AI to create quotes, facts or comments that did not actually occur.

We do not pass off generated images as real photographs.

We do not use AI to create or alter photos, videos or audio recordings of real people and events in a way that could mislead our audience, nor do we synthesise the voices of real people.

IV. Labelling

Machine translation that has not yet been proofread by a human is indicated beneath the headline.

Images or videos created or altered by generative AI are captioned to specify the tool used and the real-world materials on which they are based, for example: «Please note: This is not a real photograph. Illustration generated using Adobe Firefly based on…». The label is also placed on the image itself.

Technical processing of photos using AI is not labelled: increasing resolution, reducing noise, colour correction and cropping to change the aspect ratio (for example, from portrait to landscape). The condition for cropping is that the added section must contain only a neutral background and must not include people, objects, text or details that affect the content of the image. If this condition is not met, the photo is labelled as having been altered using AI.

V. Confidentiality

The editorial team distinguishes between two types of AI services. Tools integrated into the editorial systems at the discretion of the media director (machine translation, the IMS tool, the in-house assistant) may be used for drafts and published material. It is prohibited to upload unpublished material to publicly available AI services without the editor’s approval.

It is prohibited to upload data from anonymous sources, documents received from informants on a confidential basis, personal data of individuals featured in publications, as well as passwords and access keys to any AI services.

VI. Responsibility

The decision to integrate AI tools into the editorial systems is made by the media director, Oleg Derenyuga. Editor-in-Chief Kateryna Sereda is responsible for ensuring compliance with this policy in content. Errors resulting from the use of AI are corrected in accordance with the rules set out in the «Corrections» section. This policy is reviewed twice a year alongside the editorial policy.