Google Now Calls AI Answer Manipulation Spam. Here Is What That Means For Your GEO Strategy

Google has confirmed that its spam policies extend to attempts to manipulate AI-generated answers in Search, and the June 2026 spam update is already enforcing that rule. For any business investing in AI SEO, also known as GEO or generative engine optimisation, this is worth understanding properly, because the line between legitimate optimisation and penalised manipulation has just become a lot more visible, even if it is still not entirely clear where it sits.

Why AI Answers Are Easier To Influence Than You Might Think

A recent Cornell Tech research paper looked at how AI research tools actually gather their sources. When a tool answers a question, it does not read one page and summarise it. It fires off a batch of related sub-queries, retrieves whichever pages keep reappearing across those queries, and stitches together a report from the most frequently surfaced sources.

The researchers found that this process leans heavily on community pages such as Reddit threads and forum discussions. Within a single topic cluster, one user-generated page turned up in as many as 48 percent of the sub-queries, and user-generated platforms accounted for somewhere between 17 and 23 percent of every URL retrieved. That concentration is the vulnerability. If a handful of pages get pulled repeatedly, then whatever sits on those pages carries disproportionate weight in the final answer.

The researchers tested this directly. They found that around 13 words of planted text on a recurring page was enough to insert a chosen brand or product into the finished AI report in 38 to 51 percent of the sessions that retrieved that page. Spreading similar text across several pages pushed that figure up to between 42 and 62 percent. Even when the planted text made up less than 4 percent of a long page, it still influenced the output in 30 to 53 percent of cases.

This is not a small loophole. It is a structural feature of how these research agents work, and it is part of why Google has now drawn a clear line around it.

What Counts As Manipulation

Google’s existing spam policies already covered things like keyword stuffing, cloaking, and link schemes. The update clarifies that attempts to manipulate generative AI responses sit in the same category. In practice, that means planting brief, unattributed recommendations on forums or review sites with the specific intention of nudging what an AI tool says, rather than genuinely earning that mention through real customer experience, is treated as spam.

Google’s own documentation backs this up directly. The company’s official Search Essentials spam policy page now defines spam as any technique used to deceive users or manipulate Google’s systems into featuring content prominently, and it specifically names attempting to manipulate generative AI responses alongside the older practice of manipulating rankings. That single sentence is the policy basis behind everything described above, and it confirms this isn’t an informal stance buried in a blog post. It’s written directly into Google’s core guidance for what counts as spam in Search.

The honest answer is that nobody, including Google, has drawn a precise boundary yet. The tactics that genuinely lift a brand into AI answers and the tactics that count as manipulation look similar from the outside. The difference comes down to intent and authenticity rather than the mechanics of where the content sits.

What This Means For Your Business

The takeaway is not to avoid community platforms. It is to be deliberate and genuine about how you show up on them.

Identify where your category actually gets discussed. For most SMEs that means specific subreddits, local Facebook groups, trade forums, or review platforms relevant to the service or product. These are the pages AI tools keep returning to, so a credible presence there matters more than another blog post on your own site.

Earn mentions rather than plant them. Real customers leaving honest reviews, genuine participation in community discussions, and answering questions as a knowledgeable practitioner all build the kind of presence that holds up under scrutiny. Coordinated or anonymous text designed to look organic is exactly what the new policy targets, and it is increasingly detectable.

Concentrate effort rather than spreading it thin. The research suggests that a small number of well-placed, credible mentions on the right high-traffic community pages outperform a scattershot approach across dozens of low-relevance ones.

Monitor manually, because there is currently no dashboard that tells a business whether it appeared in an AI answer, was cited in a generated report, or was passed over entirely. Spot checking AI Overviews and AI Mode results for the queries that matter to your business is, for now, the only way to see where you stand.

The Bigger Picture

AI visibility is becoming a surface that needs active monitoring rather than a channel you set up once and leave alone. As more search queries get answered directly inside Google’s AI features rather than through a list of links, the businesses that show up will increasingly be the ones with a genuine, well-established presence across the community platforms these tools rely on, not the ones that found a clever workaround.

If you want a clearer picture of where your business currently stands in AI-generated answers or want help building an authentic GEO strategy that holds up under Google’s tightened spam policy, get in touch with the team at Clarus Digital.