Campaigns Are Optimizing Their Websites for Chatbots
Matthew Leo · Published October 31, 2026 · AI
Political campaigns have spent years optimizing pages for Google. A newly reported U.S. campaign proposal shows the next target: the answers produced by AI chatbots.
Reuters reviewed a pitch prepared for Republican candidate Mike Minogue’s campaign against Massachusetts Governor Maura Healey. The proposal described publishing hundreds of policy pages, earning placements on sites chatbots tend to cite, using Reddit and YouTube, and monitoring how models answer questions about the candidates.
The campaign did not comment to Reuters, so the document is evidence of a proposed strategy, not proof that Minogue adopted or executed it. That distinction is important. The pitch is still valuable because it shows how political consultants think generative search can be influenced.
Generative optimization is not ordinary search optimization
Traditional search optimization tries to make a page rank prominently in a list of links. Generative engine optimization, often shortened to GEO, tries to make a person, claim or framing appear in a chatbot’s synthesized answer.
The tactics overlap: clear pages, structured facts, reputable links and repeated language can help both. The difference is the interface. Search results let a voter see competing sources. A chatbot may compress them into one fluent paragraph, obscuring which facts came from the candidate, an independent report or a partisan intermediary.
That creates three distinct risks:
- Source laundering: A campaign claim can sound independent after a model blends several pages.
- Scale: Hundreds of narrowly targeted pages can cover more questions than reporters or opponents can easily audit.
- Instability: Model answers change with wording, updates and retrieval sources, making campaign promises about “winning” an answer hard to verify.
Canadian law does not have a chatbot-specific rule
Elections Canada’s digital-literacy guidance warns that chatbots can confuse levels of government and electoral systems, and that answers may be incomplete or inaccurate. Its recommendations on AI and elections call for greater transparency and for chatbots to direct people to authoritative voting information.
The Canada Elections Act does not specifically regulate artificial intelligence as a category. Existing rules can still apply, including prohibitions on falsely attributing material to a candidate or election authority. A tactic can also be legal while remaining misleading or difficult to audit.
Canadian campaigns and platforms should therefore disclose who paid for pages designed to influence automated answers, keep policy claims on stable official URLs and preserve change histories. Chatbot providers should surface citations prominently and distinguish first-party campaign material from independent sources.
A four-step check for voters
- Ask a factual question. “What is the candidate’s housing policy?” is easier to verify than “Who has the best plan?”
- Open every citation. Check whether it is the candidate’s site, an election authority, a news report or an anonymous post.
- Find the primary document. Read the platform, voting rule or financial filing instead of relying on the summary.
- Cross-check the jurisdiction. Canadian federal, provincial and municipal responsibilities are often mixed up in generic answers.
Use Elections Canada or the relevant provincial election agency for registration, identification, polling locations and deadlines. A chatbot can help locate information, but it should not be the final authority on how to vote.
The deeper issue is not that campaigns will publish persuasive material; that is what campaigns do. It is that a synthesized answer can hide the line between persuasion and verification. Voters need that line made more visible, not smoother.
Tags: AI governance, elections, AI assistants, Canada, misinformation