Google has widened its Search spam policies so they now explicitly cover attempts to manipulate generative AI responses in Google Search. That sounds like a technical rule for website owners, but it matters for ordinary users too: more people are now seeing AI-written summaries before they ever click a traditional search result.
The change was spotted by Search Engine Land and reported by The Verge. Google’s own policy page now says spam includes techniques used to deceive users or manipulate Search systems, including attempts to manipulate generative AI responses in Google Search. Sites that break the rules may rank lower or disappear from results altogether.
In plain English, Google is saying that the old search-spam problem has moved into the AI era. It is no longer only about stuffing pages with keywords or building throwaway sites to rank highly. The same incentive now exists around AI answers: if a brand, publisher, scammer or affiliate site can persuade an AI system to mention them, recommend them or treat them as unusually authoritative, that mention may shape what millions of people believe.
Why this matters when you use AI search
AI search can be genuinely useful. It can summarise complicated pages, compare options, explain unfamiliar terms and save you from opening twenty tabs just to understand the basics. But it also changes the trust problem. With normal search results, you can at least see a list of pages and choose which source to open. With an AI summary, the answer may feel more settled than it really is.
That is why manipulation matters. If a search answer is partly shaped by pages that were written mainly to influence an AI system, the result may look helpful while quietly reflecting someone else’s commercial agenda. This is especially awkward for everyday topics such as shopping, travel, home repairs, health information, financial decisions, software recommendations and local services.
ManyHands has covered this wider shift before, including when Google’s live AI search helps and when normal search is better. The useful rule has not really changed: AI search is best treated as a starting point, not the final word.
What “manipulating AI search” can look like
Some of it may look familiar. A page might be stuffed with unnatural phrases, built around fake comparisons, or dressed up as independent advice when it is mainly there to push one product. Other tactics are newer. The Verge points to examples such as recommendation poisoning, where content is written to nudge large language models into remembering or repeating a particular site as authoritative.
There is also a growing industry around “generative engine optimisation”, often shortened to GEO. Some of it will simply mean making content clear, useful and easy for AI systems to understand. Fair enough. But the risk is obvious: once AI answers become valuable real estate, some people will try to game them. The internet, displaying its usual restraint. Naturally.
For UK users, the practical issue is not whether every AI answer is suspect. It is whether you can spot the moments when an answer deserves a second look. If the AI summary recommends a product, names a “best” provider, ranks services, gives health guidance or suggests a financial step, that is when the source trail matters.
Three checks before trusting an AI search answer
First, look for named sources. A useful AI answer should make it reasonably easy to see where the claim came from. If the answer gives a strong recommendation but the sources are thin, promotional, out of date or all suspiciously similar, treat that as a warning sign.
Second, click through when the decision matters. For low-stakes questions, an AI summary may be enough. For anything involving money, health, safety, legal rights, work, children or a purchase you would regret, open the underlying pages. Check whether the source is official, editorial, commercial, or just another rewritten article repeating the same claim.
Third, compare the AI answer with ordinary search results. If the AI summary says one thing but the visible results tell a more mixed story, do not assume the summary has magically resolved the debate. It may have simplified too aggressively, missed newer information, or leaned on sources that look confident but are not especially trustworthy.
What this means for website owners and small teams
Google’s update is also a useful nudge for anyone running a blog, business site, charity site or side project. Trying to trick AI systems may bring short-term attention, but it could also make a site look spammy to Google. Clear, original, accurate content remains the safer long-term bet.
That does not mean ignoring AI search. It means writing pages that answer real questions, explain who is behind the information, show dates where they matter, and avoid pretending to be independent when the page has a commercial reason to exist. For small organisations, that is less glamorous than “AI optimisation”, but it is also less likely to age like milk.
It also connects with a broader safety point. We recently wrote about AI tools ignoring instructions and why permissions need checking. Search manipulation is a different problem, but the habit is similar: do not hand full trust to an automated system just because it sounds fluent.
The takeaway
Google’s policy update is a sign that AI search is becoming important enough to attract the same old web-spam games. The company says it can use automated systems and human review, and sites that violate its spam policies may be pushed down or removed. That is useful, but it does not remove the need for human judgement.
For everyday users, the sensible approach is simple. Use AI search to get oriented. Ask follow-up questions. But when the answer affects your money, health, work, safety or family, check the sources yourself. The AI may be helpful. The web behind it may be messy. Both things can be true.
Sources: The Verge; Google Search spam policies.
