For Politicians and Elected Officials
Cognitive Sovereignty Checklist for Politicians and Elected Officials
AI tools can write eloquent speeches and personalise constituent messages at scale, but they cannot know your community, your values, or what you actually believe in. When you use AI to shape policy positions or author communications without deliberation, your constituents eventually detect the inauthenticity and your electoral trust erodes. This checklist helps you use AI as a tool for research and drafting while keeping your own judgement at the centre of your work.
These are suggestions. Take what fits, leave the rest.
0 / 19 applicable
Tap once to check, again to mark N/A, again to reset.
Before You Use AI for Speeches and Public Statements
Write your core message in your own words firstbeginner
Before opening ChatGPT or Claude, spend fifteen minutes writing the argument you actually want to make. This forces you to do the deliberative work yourself and gives AI something authentic to improve rather than something hollow to polish.
Test the speech against your voting record and past positionsbeginner
Read the AI-drafted speech and ask whether it contradicts something you have said or done in the past two years. If your constituents can fact-check you into inconsistency, the eloquence of the language will not save you.
Identify which claims require local knowledge to defendintermediate
AI knows what is written about your region online. It does not know what local business owners told you last month or what the real unemployment rate feels like in your poorest ward. Mark those claims and prepare your own evidence or testimony for them.
Practise the speech aloud before delivering itbeginner
If you cannot speak the words naturally or remember the argument under pressure, you have not made it your own. Practising forces you to internalise the logic so you can defend it in a follow-up question.
Remove phrases that sound unlike you when you hear them aloudintermediate
AI-generated language often uses word choices and rhythms that differ from your natural speech. Your constituents will hear the difference and question whether you wrote or understood what you said.
Check that the speech reflects your actual position on the main issueintermediate
AI can synthesise what multiple sources say about a topic and produce something that sounds balanced but commits to nothing. If the speech allows you to claim victory to different audiences on the same question, it is not authentic representation.
When Using AI for Policy Research and Position Formation
Separate research gathering from position makingbeginner
Use Perplexity or Claude to gather evidence on a policy question. But do not ask AI to recommend a position. Read the evidence yourself, talk to advisers who know your community, and form your own view before you commit publicly.
Verify that AI summaries of opposing views are fairintermediate
AI tends to present opposing arguments in their weakest form because it is trained to avoid genuine conflict. If you cannot articulate the best case for the position you disagree with, you do not understand the issue well enough to represent your constituents on it.
Ask your policy adviser to challenge any position AI suggestsintermediate
Before you adopt a position that AI analysis has shaped, have one person whose job is to argue against it. This forces you to defend your thinking and catches cases where you have accepted plausible-sounding policy that does not suit your community.
Document which parts of your policy position came from your own thinkingadvanced
If a journalist or opponent asks where you stand on an issue, you must be able to answer from memory and conviction, not by checking what you asked AI to produce. You should know your own policy positions without consulting your prompt history.
Require AI research summaries to cite their sources by namebeginner
Do not accept a summary that says 'research shows' without the specific study, publication, and date. You need to be able to verify the evidence yourself and check whether it applies to your context.
Notice when AI is synthesising a consensus that does not actually existadvanced
AI can make marginal or contested positions sound settled because it averages across its training data. Before you commit to a policy position, verify that credible experts actually agree on it in the real world.
Consult with community leaders before adopting AI-researched positionsintermediate
Policy analysis from AI does not include knowledge of your specific community's needs or readiness for change. Talk to school heads, business owners, and people in affected sectors before you public ally adopt a position that AI analysis has shaped.
When Using AI for Constituent Communication and Relationship Building
Write personal responses to significant constituent concerns yourselfbeginner
When a constituent has invested effort in a detailed letter or raised a serious concern about a local issue, your response should come from you. AI-personalised template responses will feel hollow to someone who made themselves vulnerable by writing to you.
Use AI to draft routine acknowledgements only, not substantive repliesbeginner
AI can help you quickly acknowledge receipt of a constituent message or handle scheduling questions. It should not write your actual answer to a policy question or complaint about local services.
When AI personalises messages at scale, review a sample of them for authenticityintermediate
Some tools promise to personalise constituent communications by pulling details from public records or social media. Read five examples before you deploy this at scale to your voter base. If the personalisation feels creepy or false to you, it will feel that way to your constituents too.
Tell constituents when you are using AI assistance in your responseadvanced
If a constituent receives a message that was drafted or shaped by AI, they should know it. Hiding the use of AI is a form of inauthenticity that will damage trust if discovered later.
Keep a record of which communications were AI-draftedintermediate
If a journalist, opponent, or your own party asks whether a constituent message you sent was AI-written, you need to be able to answer truthfully from your records. Ambiguity about authorship will become a liability.
Never use AI to manufacture the appearance of personal engagementbeginner
Do not use AI tools to send dozens of personalised messages that suggest you have been personally reading and responding to constituent mail when you have not. Your constituents will eventually learn the truth and will feel manipulated.
Five things worth remembering
- Your authenticity is your brand in politics. If voters believe you have outsourced your thinking to AI, they will vote for someone they believe is actually thinking.
- Always ask yourself: could I defend this speech, policy position, or constituent reply in a live debate where I cannot check my notes? If the answer is no, it is not authentic enough to send.
- When AI produces something eloquent that does not sound like you, that is a warning signal. Rewrite it until it uses your words and your rhythm.
- Keep a small group of trusted advisers who know your community and ask them to read AI-drafted work before it goes public. They will catch inauthenticity faster than you will.
- If you cannot explain your own policy position in simple language without consulting your AI notes, you do not understand it well enough to represent it in parliament or your council.
Related reads
Common questions
Should politicians and elected officialss write your core message in your own words first?
Before opening ChatGPT or Claude, spend fifteen minutes writing the argument you actually want to make. This forces you to do the deliberative work yourself and gives AI something authentic to improve rather than something hollow to polish.
Should politicians and elected officialss test the speech against your voting record and past positions?
Read the AI-drafted speech and ask whether it contradicts something you have said or done in the past two years. If your constituents can fact-check you into inconsistency, the eloquence of the language will not save you.
Should politicians and elected officialss identify which claims require local knowledge to defend?
AI knows what is written about your region online. It does not know what local business owners told you last month or what the real unemployment rate feels like in your poorest ward. Mark those claims and prepare your own evidence or testimony for them.