GEO
The Google AI Spam Policy Rewards Real Expertise. Here Is What to Do About It.
The Google AI spam policy now treats attempts to manipulate AI answers as spam, and the marketer coverage this week makes the message hard to miss. Google expanded its Search spam policy so that gaming its generative AI responses, the AI Overviews and AI Mode that run on Gemini, can trigger the same demotion or removal that ranking manipulation always has (Google Search Central). The tactics being called out are the shortcuts: low-value pages built to anticipate queries and fake FAQ content written to catch questions people might ask. My view is simple. Visibility in AI answers cannot be manufactured, and the brands that win will be the ones with real expertise, their own data, and a point of view worth citing.
What changed in the Google AI spam policy?
Google rewrote the opening definition of its spam policy. The line now reads that spam refers to techniques used to deceive users or manipulate Search systems into featuring content prominently, "such as attempting to manipulate Search systems into ranking content highly or attempting to manipulate generative AI responses in Google Search." The earlier version stopped at ranking content highly (Search Engine Land). The addition is small in word count and large in scope. Every tactic Google already treats as spam now applies when the target is an AI answer, across AI Overviews and AI Mode.
The practical read for marketing leaders arrived with this week's coverage aimed at CMOs, which framed the update as a direct warning to anyone optimizing for AI citations (Forbes CMO Network). There is no separate review track for AI-answer manipulation and no exemption for tactics dressed up as legitimate optimization. The enforcement mechanism is the one Google has used for years, so the penalties are familiar and real.
What tactics does the AI spam policy target?
The update names the shortcuts teams have been tempted to try. Coverage of the change highlighted generating low-value pages designed to anticipate the queries a user might type, and writing fake FAQ content built to match questions in the hope of surfacing in an AI answer (Forbes CMO Network). These join tactics Google already polices: site reputation abuse, mass-produced pages with little value, synthetic authority signals, and manufactured mention patterns meant to steer AI-generated responses.
What connects all of them is intent. Each one exists to influence a machine's summary without adding anything a person would value. That is the specific behavior Google widened the policy to catch, and it is the same behavior that damages a brand when a customer actually reads the page. The tactic and the trust problem are one and the same, which is why this is not only an SEO concern.
Why can't AI search visibility be built on shortcuts?
Shortcut content now carries two costs that compound. The first is enforcement. A page or a whole site can be demoted or removed, and the volume play that produced dozens of thin pages becomes a liability the moment Google acts on it. The second cost is reputational. Fabricated FAQs and pages written for an algorithm are, by definition, pages your customers may land on, and they represent your brand poorly when they do.
There is a strategic reason to care beyond compliance. Even setting Google aside, the non-Google engines that also shape discovery, including ChatGPT and Perplexity, set their own rules and weight credibility signals in ways no single tactic can reliably game. Building visibility on shortcuts means building it on ground that shifts every time one platform updates. Building it on genuine expertise means the same asset works across Google's AI answers, the other engines, and the human reader at the end of the chain. One is fragile. The other holds. For the broader playbook on getting named in AI answers, see my 2026 guide to generative engine optimization.
What should marketing and comms teams do instead?
This is where I want to be direct, because the answer is not comfortable and it is not fast. You do need real experts. You do need your own data. You do need your own insights. There is no version of durable AI visibility that skips those three, and every shortcut on offer is a way of pretending otherwise.
Credible authorship is the first signal. Real, named people with verifiable credentials should carry your most important content, with linked bios and a consistent identity across the web, because expertise attached to a person is far harder to fake than a page written by no one. Original data is the second. Proprietary research, your own benchmarks, and numbers only your organization can produce give AI systems and readers something they cannot get anywhere else, which is exactly what earns a citation. A defensible point of view is the third. A clear, argued position that reflects genuine experience is the one thing a competitor cannot copy and a model cannot synthesize from the average of everyone else's content.
Consistency ties them together. Your facts, your positioning, and your claims should match across your site, your profiles, and your earned coverage, because AI systems reward sources that agree with themselves and lose confidence in ones that do not. Fewer, stronger pages beat a large library of thin ones, so retiring low-value content is a real part of the work, not an afterthought.
What does a defensible authority strategy look like in practice?
Start by auditing what you already have. Identify the thin, algorithm-first pages that could represent a trust risk and retire or consolidate them, then count how many of your important pages carry a named, credentialed author. Most teams find the number is low. Fix that next, by pairing your genuine subject-matter experts with the content that matters and giving them real bylines.
Then commit to producing something the internet does not already have. One original survey, one proprietary benchmark, or one honest analysis of your own client results is worth more to AI visibility than fifty pages assembled from what everyone else already published. Route your content strategy toward authority building with named experts, partners, and original research, and treat volume as the wrong target. For comms teams specifically, this is another reason to align earned media and thought leadership with the same experts and the same data, so the brand says one consistent, credible thing everywhere it appears.
Give the work an owner and a cadence. Assign each pillar topic to a named expert who is accountable for the point of view on it, and put a standing slot on the calendar for original data, whether that is a quarterly benchmark, a recurring customer survey, or a written teardown of a real engagement. The teams that produce citable material consistently are the ones that scheduled it, not the ones that waited for inspiration. Measure the right thing too. Track how often your brand and your named experts are cited in AI answers and earned coverage, not how many pages you shipped, because citation is the outcome this policy change rewards and page count is the habit it penalizes. If you want to instrument that, my guide to prompt monitoring tools covers how to track your citations across engines.
None of this is a reason to abandon AI tools in your workflow. The problem Google named is narrow: using AI to mass-produce empty pages aimed at a machine. Using AI to help produce genuinely useful content was never the target. Use the tools to draft faster, to structure research, and to pressure-test an argument, then put a real expert's judgment, a real dataset, and a real position on top. That combination is both allowed and effective, and it is the version of scale that does not carry trust risk.
In my own client work, the pages that consistently earn AI citations are the ones built on data the client owns and a named expert willing to take a position. The generic, high-volume pages rarely get cited, and after this policy change they carry a risk they never justified in the first place.
The move for the second half of 2026
Google drew a clearer line this week, and the line favors the brands that were already doing the harder work. Treat this as permission to stop competing on volume and start competing on expertise. Put your real experts forward, publish the data only you have, and say something you actually believe. That is the strategy that survives the next policy update, and the one after that, because it was never a tactic to begin with.
This policy change rewards the brands that stopped competing on volume and started competing on expertise. Want help auditing your content for trust risk and building an authority-first plan? Text Alyssa.
“Text” Alyssa