For Journalistss and Reporters

Cognitive Sovereignty Checklist for Journalistss

About 20 minutes Last reviewed March 2026

AI tools like ChatGPT and Perplexity can replace your most valuable journalistic instinct without you noticing. When you let AI summarise your research, transcribe your interviews, or shape your story structure, you stop engaging directly with the material that teaches you what actually matters. Your news judgement depends on this friction. Protect it.

Cognitive sovereignty insight for Journalists: a typographic card from Steve Raju

These are suggestions. Take what fits, leave the rest.

Download printable PDF

0 / 19 applicable

Tap once to check, again to mark N/A, again to reset.

Protect Your Source Relationships

Conduct interviews yourself instead of asking AI to summarise thembeginner

When you listen to a full interview or read a full transcript, you hear the hesitations, contradictions, and moments where someone chooses their words carefully. AI summaries flatten these signals. Your relationship with sources also depends on you doing the work of understanding them.

Read primary sources in full before asking AI to summarise thembeginner

A study, report, or court filing contains detail that shapes how you should read it. When you skip to an AI summary, you miss the context that tells you whether the summary is emphasising the right things or missing what actually matters for your story.

Schedule a source call instead of emailing questions to verify AI researchbeginner

Email verification is faster, but phone conversations reveal whether a source actually understands what they are saying, how confident they are, and what they might be leaving out. These conversations also build the trust that makes sources more likely to talk to you about harder stories later.

Keep a list of sources you have talked to for each storyintermediate

When the list is short, the story is probably built on AI-summarised research rather than your own reporting. Useful to know before you file, not after. It also shows which angles still need a real source.

Ask at least one source to explain something you think you already understandintermediate

This is how you discover what you are missing. A source may tell you that everyone in their field knows something different from what the literature says, or that a policy sounds good on paper but works differently in practice. AI research will never catch this.

Develop sources before you need them for a storyadvanced

When you only talk to people after AI has suggested they might be relevant, you are reacting to the summary instead of building genuine reporting relationships. Sources you have built trust with over time are more willing to share difficult information.

Preserve Your News Judgement

Write your own story structure before showing a draft to AIbeginner

If you ask AI to suggest how to organise your story, you are outsourcing the decision about what matters most and why. Your structure should reflect your reporting and your sense of what the reader needs to understand first.

Note what made you decide this story matters before you start reportingbeginner

Write one paragraph about why this story is important and for whom. Return to this as you report. If your thinking changes, notice that. If AI tools are pushing you toward different angles, ask yourself whether the AI is right or whether it is just optimising for engagement metrics.

Resist using Perplexity or similar tools as your first reporting stepintermediate

These tools are designed to answer questions quickly. But reporting is about finding questions that matter. If you use AI to define the territory before you have your own instinct about it, you will report within the boundaries AI has already set.

Track which story ideas came from your own reporting versus AI suggestionsintermediate

Over time this shows you whether you are developing your own sense of what matters or following AI-generated angles. Younger journalists especially need to build their instinct before they can judge which AI suggestions are valuable.

Ask sources what surprised them about the story you are working onintermediate

Sources often know what journalists usually get wrong. This is where your reporting diverges from the AI-summarised conventional wisdom. These moments are where real news judgement lives.

Read your competitors' coverage, then ask yourself what they missedadvanced

This is different from asking AI what the story should be. You are training your own sense of what matters. If you cannot answer this question without AI help, that is a sign you need more direct source engagement.

Spend time reporting on stories you think will not workadvanced

This is how you develop the instinct to recognise when a story is worth pursuing. AI tools will always optimise toward stories that seem to have good engagement potential. Your judgement comes from knowing the stories that matter even when they do not look like they will be popular.

Verify and Ground Your Work

Use Otter.ai for transcription only, not for note-takingbeginner

Transcription is a tool. But if you rely on it to capture what an interview meant, you are missing the moment-by-moment judgement that comes from listening and taking notes yourself. Your notes capture what struck you as important.

Check every factual claim in AI-written first drafts against your reporting notesbeginner

AI can construct sentences that sound true but rest on facts that are slightly wrong or taken out of context. Your reporting notes show whether the AI version matches what you actually learned.

Do not use AI summaries of studies as a substitute for reading the methodology sectionbeginner

The methodology is where you learn what the study actually tested and what it cannot tell you. An AI summary may describe findings without explaining the limits. This is how accurate-but-false stories get built.

Keep a running document of facts that AI got wrong in your draftsintermediate

Over time this teaches you which kinds of facts AI is likely to mishandle. You will start fact-checking these categories more rigorously instead of trusting the AI output.

Read the source material yourself before you read an AI summary of itintermediate

This trains your eye to notice what the AI missed or emphasised. You will develop a sense of when AI summaries are reliable and when they are flattening important complexity.

Ask a source to verify an AI-generated quote paraphrase before you publish itintermediate

If you have written a paraphrase of what a source said and an AI tool helped you refine it, the source should confirm that your version captures their meaning. This protects you and keeps the source relationship honest.

Five things worth remembering

The first draft from AI often feels finished enough. It is not. Your real reporting work is recognising what the AI draft is missing because it did not do the reporting you did.
When you feel like an AI tool is saving you time on research, ask yourself whether you are really saving time or just deferring the work to the moment when a source contradicts your AI-summarised version.
Younger journalists should use AI for transcription and first drafts, but spend at least half their reporting time away from the tools. This is how they learn what a story actually is.
Your news judgement is not a permanent skill. It atrophies when you stop doing the work that builds it. Every story where you skip source engagement is a story where you do not practise the instinct you need.
A story built on AI-summarised research can have every fact correct and still miss what the facts mean. This is the trap. Check not just whether your facts are right, but whether your story explains why anyone should care.

Prompt Pack

Paste any of these into Claude or ChatGPT to pressure-test your own judgment. They work best when you respond honestly before reading the AI reply.

Form your editorial angle before AI input

I am working on a story about [topic]. Before I use any AI research tools or consult existing coverage, ask me questions about what I know, what I think the real story is, and what I am going to do to find out. Help me articulate my own editorial instinct first.

Test an AI-generated summary against your judgment

I have used AI to summarise background research for a story. Ask me questions that reveal whether I actually understand the primary sources and context, or whether I am about to report a story I only understand through AI intermediation.

Find the angle AI research missed

I have done AI-assisted background research on [topic]. Now act as a veteran editor: what questions would you ask me that the AI research almost certainly did not answer? What angles does an algorithm trained on existing coverage systematically miss?

Rebuild your interview preparation instinct

I have an interview coming up with [describe source]. Help me prepare. But do it by asking me questions rather than generating a list of questions for me. I want to develop my own line of inquiry before you offer yours.

Examine your source relationship quality

I want to examine how AI tools have changed my source relationships and reporting habits. Ask me questions about how I build source trust, how I conduct interviews, and how much of my current story research involves direct human contact versus AI-mediated information gathering.

Reading List

Five books that give this topic the depth it deserves. Each one is genuinely worth reading, not just citing.

The Elements of Journalism

Bill Kovach and Tom Rosenstiel

The foundational account of what journalism is for and what it requires, a useful anchor when AI tools push toward efficiency over accountability.

Flat Earth News

Nick Davies

A rigorous examination of how journalism fails when reporters stop doing original reporting. The AI equivalent of churnalism is already here and this book gives you the framework to recognise it.

Thinking, Fast and Slow

Daniel Kahneman

The cognitive biases that make AI-generated information so easy to accept without scrutiny. And that good editors have always tried to counteract.

Stolen Focus

Johann Hari

The conditions required for deep investigative thinking, sustained attention, slow reading, relationship-building. And what digital tools, including AI, do to those conditions.

Cognitive Sovereignty

Steve Raju

A framework for protecting independent editorial judgment as AI tools reshape how journalists research, write, and find their angle.

Questions to ask yourself

Use these before your next AI-assisted decision. Honest answers are more useful than comfortable ones.

When did I last pursue a story angle that came entirely from my own observation or source relationship, not from AI research?
Is my editorial instinct. The nose for what matters and what does not, getting stronger or weaker?
How much of my current story understanding is based on direct source contact versus AI-mediated information?
Could I write this piece without AI tools? Is it still genuinely mine?
Am I developing the source relationships that make journalism possible, or am I becoming a skilful user of information intermediaries?

Common questions

Should journalists use AI writing tools?

AI tools are useful for transcription, initial research sweeps, and drafting routine structured content. But journalism depends on source relationships, the judgment to know what matters and what is misleading, and the ability to read what a source is not saying as much as what they are. None of that is accelerated by AI. And the parts of reporting that AI handles well are also the parts most likely to carry errors that look plausible.

What are the risks of AI-generated journalism?

The primary risks are confident inaccuracy, AI tools present fabricated information with the same tone as verified information. And the homogenisation of perspective. AI trained on existing media reproduces the angles, framings, and assumptions already present in journalism rather than challenging them. The most important journalism tends to be exactly the kind that cuts against consensus.

Can AI replace investigative journalists?

No. Investigative journalism requires building trust with sources over time, reading situations and people, exercising judgment about what to pursue and what to protect, and the ethical reasoning to handle sensitive information responsibly. AI can help analyse large datasets in investigative work. But the reporter's judgment about what those patterns mean and whether to publish remains entirely human.

How is AI changing journalism?

AI is automating routine reporting, financial results, sports scores, structured data stories. And accelerating research tasks. It is also creating significant problems: AI-generated misinformation that looks credible, deepfakes, and the devaluation of commodity content that fills most news sites. The journalism that AI cannot do is relationship-based, judgment-dependent, and willing to take the kind of editorial risk that an algorithmic tool trained on what people already believe cannot take.

How can journalists maintain their independent voice when using AI tools?

By writing their own first draft before consulting any AI tool. By interviewing sources directly rather than using AI to summarise press materials. By actively questioning what angle they would have taken before AI suggested one. The editorial instinct is built by the decisions you make, not the decisions AI makes that you approve.