AI seeding: How brands shape AI answers
Think about the last time you asked AI for a recommendation. The headphones, the laptop, the restaurant for Friday night. It felt like an honest answer.
But what if that answer was engineered?
The restaurant that doesn't exist
In May 2026, Cornell Tech researchers made up a restaurant, added one short recommendation for it to a Reddit discussion, and asked AI research tools for the best Mexican restaurants near Austin. The AI recommended it (Help Net Security).
Just 13 words on a single page were enough to appear in 38–51% of the AI reports tested (Cornell Tech study). It was a controlled test on open-source tools, not live ChatGPT, but it shows how much AI trusts what people write online.
Why marketers are paying attention
A June 2025 Semrush study found Reddit was cited in 40.1% of AI answers (DesignRush). AI trusts it because it's real people giving honest opinions, and OpenAI and Google both pay Reddit for its content (The Next Web).
That's created a new field: GEO, or generative engine optimisation. SEO got you to the top of Google. GEO gets you recommended by AI.
Meet AI seeding
AI seeding means placing your brand where AI learns: helpful blogs, Q&A answers, industry articles and genuine reviews. Done right, it's simply good marketing.
Some brands fake it, posting reviews from accounts pretending to be real users. But AI is fighting back. Reddit says it now catches about 25,000 spammy posts a day (iTechPost).
How to check an AI answer
Ask for its sources. Most AI search tools will list the pages they used.
Click the thread. Read the full discussion, not just the top comment.
Check who's talking. A brand-new account, or one that only praises one product, is a red flag.
Look for real agreement. Several independent people, including what they didn't like, beat one glowing post.
Good news for small brands
You don't need a big ad budget to be recommended by AI. If real customers talk about you, AI notices. At ScaleX, that's how we approach AI visibility: real customers, real conversations.
Because AI doesn't recommend the best product. It recommends the one people talk about most. And getting people talking? That's just marketing.
