Google has moved to extend its Search spam policies to cover AI-generated answers, a policy shift that carries direct consequences for Australian businesses and the regulators already scrutinizing the reliability of AI-generated information. The trigger was a BBC investigation that demonstrated how easily Google's Overviews and AI Mode features could be fed false information from planted web content.
The vulnerability was illustrated by BBC Future journalist Thomas Germain, who spent roughly 20 minutes writing a fabricated listicle on his personal website. Within 24 hours, Google's Gemini app, AI Overviews, and ChatGPT were repeating the invented claim as fact. No human review step intercepted the round trip from planted content to AI-generated answer.
For Australian enterprises that depend on search visibility for brand exposure and customer acquisition, the policy change exposed a structural weakness in the AI search layer they already rely on.
The May policy update places attempts to game Google's AI search features on the same enforcement footing as conventional ranking spam. The update covers AI Overviews and AI Mode, and operators using tactics such as biased ranking listicles and prompt-injection patterns face demotion or removal from results.
The company also says the revised language clarifies rules it already enforced, rather than introducing new ones.
What this means for Australian enterprises
For Australian businesses, the manipulation problem represents a structural shift in the commercial stakes of search visibility. Under traditional search, a biased or low-quality page might rank well but still compete against other visible links. AI search collapses that dynamic: a manipulated answer can reach users as a settled, uncontested response, with research suggesting readers are roughly 58% less likely to click through to underlying sources when an AI Overview is present.
The risk extends to information quality more broadly. AI Overviews now appear in billions of Google searches each month. An independent analysis of Google's Gemini models found accuracy rates of 85% for Gemini 2 and 91% for Gemini 3 across 4,326 tested queries each. The figures sound reassuring until scaled to Google's projected volume of over five trillion searches in 2026.
Australian enterprises that have invested in search visibility — across sectors from financial services and retail to healthcare and professional services — now face a dual challenge. They must manage their own search presence under Google's expanded spam rules while also contending with the possibility that competitor or bad-faith content could distort how their products, services, or reputations appear in AI-generated answers.
Since this policy update is new, the enforcement mechanism for AI spam remains opaque; the first meaningful signal will come when Google issues manual action notices via Search Console.
Australian regulators are already watching
The development lands at a moment when Australian regulators have been building a formal case for tighter oversight of major digital platforms and emerging AI risks.
The Australian Competition and Consumer Commission (ACCC) released the final report of its five-year Digital Platform Services Inquiry in June 2025, explicitly identifying generative AI as a source of new and evolving risks to competition and consumers. The ACCC has continued to advocate for a dedicated digital competition regime with service-specific codes of conduct targeting platforms with significant market power. In its 2026-27 enforcement priorities, the ACCC has identified manipulative conduct in digital markets as a key focus area. Such priorities also affect AI-enabled variants.
The broader regulatory framework governing AI in Australia remains distributed across multiple bodies — the ACCC, the Office of the Australian Information Commissioner, the Australian Communications and Media Authority, and the eSafety Commissioner. Each body holds a piece of the mandate without a single overarching AI Act.
Australia's AI Safety Institute, which became operational in early 2026 with AU$29.9 million, or about USD$20.9m, in government funding, is focused on risk evaluation, while proposals for mandatory guardrails for high-risk AI systems remain under policy consideration.
That fragmented regulatory landscape may make it harder to respond quickly to systemic risks to information quality arising from AI search manipulation. But it also means multiple regulatory pathways exist.
What enterprises and IT leaders should monitor
The practical question for Australian enterprises is not only whether their own practices comply with the new spam definition. It also asks whether the AI search layer these enterprises increasingly depend on for brand exposure, customer acquisition, and reputational standing is structurally sound.
Three implications are worth tracking.
- First, marketing and digital teams should audit existing content strategies for tactics — including biased listicles, self-promotional claims, or prompt-injection-formatted pages — that now fall explicitly under Google's spam definition.
- Second, enterprises in regulated industries, where AI-generated answers touching financial products, health information, or legal services carry elevated consumer risk, should assess whether current AI search outputs accurately represent their offerings.
- Third, businesses operating in sectors with strong ACCC oversight should be prepared for the regulator to reference AI search manipulation in future consumer protection proceedings.
Google's intervention establishes that the manipulation of AI-generated answers is a recognized and sanctionable problem. Whether the enforcement regime proves effective remains to be seen.
For Australian regulators and enterprise technology leaders, the more pressing signal from this episode is structural. As AI-generated answers replace ranked links as the primary interface between users and information, the reliability standards that governed traditional search are no longer sufficient. A framework built for blue links does not automatically cover a world where a single AI answer shapes what millions of users believe to be true.