Independent journalism on Jharkhand’s politics, economy and society.
AI Policy (tentative)
Our approach to artificial intelligence in journalism. This is a tentative policy document and we’re open to feedback and comments at editor@jharkhandwatch.com
Jharkhand Watch uses artificial intelligence (AI) as a tool to support journalism, research and publishing. We believe these tools can help a small newsroom work more efficiently, analyse information and explain complex issues.
But efficiency cannot come at the cost of accuracy, independence or public trust. AI does not replace reporting, verification or editorial judgement. Responsibility for everything we publish rests with the human editor.
1. How we use AI
AI tools may assist us with:
- Research and document analysis: Organising information, summarising lengthy documents and identifying claims that require further verification.
- Data journalism: Cleaning and analysing datasets, identifying patterns and anomalies, and developing charts, maps and visualisations.
- Writing and editing: Brainstorming story ideas, developing outlines, improving clarity, checking grammar and suggesting headlines or search descriptions.
- Transcription and translation: Converting interviews and recordings into text, and translating material, subject to human review.
- Production and distribution: Organising content, preparing summaries and adapting published journalism for newsletters and social media.
- Visual content: Creating illustrations or other visual assets when appropriate, provided they do not mislead readers about real people, places or events.
These are permitted uses of AI assistance, not exemptions from our editorial standards.
2. Verification comes first
AI-generated information is not evidence.
AI systems can produce false claims, fabricated references, inaccurate quotations and misleading summaries. Information suggested by these tools must be checked against reliable sources before publication.
We prioritise original documents, official data, legislation, court records, direct interviews and other relevant primary sources. Secondary reporting can provide context and corroboration.
AI tools will not be treated as the sole authority for factual claims. We will not publish an assertion merely because an AI system presents it confidently.
3. Human editorial responsibility
AI may assist in producing content, but it does not make editorial decisions.
The editor remains responsible for the accuracy, fairness, context, framing and final presentation of published work. AI-assisted material must be reviewed and edited as necessary before publication.
We will not publish AI-generated journalism without meaningful human editorial intervention and verification. We will not allow automated systems to make consequential editorial decisions without appropriate human oversight.
4. Transparency with readers
We believe readers should be able to understand how AI contributes to our journalism.
Routine assistance with grammar, editing, brainstorming or internal research does not necessarily require a disclosure on every article.
We will disclose significant AI involvement where it materially shapes the published content or its presentation. Examples may include substantially AI-generated illustrations, heavily AI-assisted summaries or automated reports.
Disclosures should explain the relevant use of AI in plain language. We will not present synthetic images, audio or video as authentic documentary evidence.
5. Privacy, sources and confidential information
Journalism often involves unpublished material, personal information and confidential sources. These require particular care.
We will not enter confidential source identities, sensitive personal information or unpublished material containing sensitive information into external AI services unless the service has been assessed as appropriate for that use and adequate safeguards are in place.
Before using an AI tool, we will consider its data-handling practices, privacy protections and terms of use. Where necessary, sensitive information should be removed or anonymised before processing.
6. Copyright and originality
We respect copyright, attribution and the intellectual property of other publishers, journalists, photographers and creators.
AI tools must not be used to plagiarise published work, reproduce protected material without an appropriate basis, or circumvent paywalls and other access restrictions.
AI-generated text, images and other assets must be reviewed for originality, accuracy and potential rights issues. We will not use AI to fabricate quotations, impersonate real people or create misleading representations of actual events.
7. Fairness, bias and manipulation
AI systems can reproduce biases or distort information. Their outputs require particular scrutiny when covering politically sensitive issues, vulnerable communities, allegations of wrongdoing or other matters where errors could cause harm.
We will assess AI-assisted work for misleading framing, omissions, unsupported conclusions and unfair representation.
Synthetic or manipulated material must not be used to deceive readers about events or evidence. AI-assisted detection of manipulated content may help an investigation, but it does not by itself establish that material is authentic or fabricated.
8. Continuous review
AI technology, its risks and its legal environment continue to evolve. Jharkhand Watch will review this policy and its newsroom practices as our tools, capabilities and publishing needs change.
Readers can contact us with questions about our use of AI, or raise concerns about a published article or visual asset through our usual feedback and correction channels.
Our commitment: We use AI to support journalism, not to substitute for evidence, independent thinking or editorial accountability.
This policy has been informed by the journalism guidance in Media Helping Media’s article, Using AI in news production. Jharkhand Watch has developed its own policy for its editorial practices and operating model.
AI Disclosure: This article is repurposed and summarised using a LLM model.
Edited and Fact-checked: Romil Javed.
Last updated: 15 October, 2026.
