YMYL in AI Search: How AI Engines Treat High-Stakes Topics

This article is about marketing and search visibility. It does not give medical, financial or legal advice.

What YMYL means in SEO

YMYL comes from Google's Search Quality Rater Guidelines, the manual Google's human evaluators use to judge the quality of search results. The current edition is dated 11 September 2025. It defines YMYL topics as those where content could significantly affect the health, financial stability or safety of people, or the welfare and well-being of society.

Google groups YMYL topics into four areas: health or safety, financial security, society (including elections and trust in public institutions), and other topics where content could cause real harm. Raters do not set rankings directly, but their judgments show Google what its ranking systems should reward.

Two details in the guidelines are worth knowing. First, YMYL is a spectrum, and Google says most topics are not YMYL at all. The guidelines contrast tsunami evacuation routes (clear YMYL), a weather forecast (possibly YMYL) and music award winners (not YMYL). Second, the more harm a topic could do, the more trust a page needs to show. For clear YMYL topics, Google treats trust as the most important part of E-E-A-T (experience, expertise, authoritativeness and trustworthiness). If you are new to how this shift plays out beyond classic rankings, our guide on where GEO and SEO differ sets out the basics.

Why YMYL matters more in AI search

In a classic results page, a searcher sees several sources, each under its own name, and decides who to trust. An AI answer changes that. The engine reads the sources, writes one summary for the searcher, and reduces the sources to small citations. On a YMYL question, the engine has effectively done the judging.

That raises the stakes in both directions. An error in an AI answer about a medication, a loan or a legal right reaches the searcher without the context a list of sources would give them, and with the authority of a single, confident reply. For brands, being left out of that reply means being left out of the comparison altogether. This is the core of answer engine optimization: when the answer is the result, the sources it trusts are the ones that matter.

How AI engines handle high-stakes questions

The clearest public evidence of how Google treats YMYL in AI features comes from what happened when its AI Overviews got health questions wrong.

In January 2026, the Guardian published an investigation that found AI Overviews giving misleading health information, including normal ranges for liver blood tests shown without accounting for factors such as age, sex or ethnicity. Experts warned that people with serious liver disease could wrongly conclude their results were normal. The investigation also flagged problems with summaries about cancer and mental health.

Google's response was targeted. As TechCrunch reported, AI Overviews were removed for searches such as "what is the normal range for liver blood tests", while slightly rephrased versions could still return a summary. A Google spokesperson said the company does not comment on individual removals and works on broad improvements, and that an internal team of clinicians had found that in many instances the information was not inaccurate and was supported by high-quality websites.

Three lessons follow for marketers in regulated categories:

• AI features on YMYL topics are watched closely and can change quickly. A visibility pattern on a health or finance query can disappear overnight, through no action of yours.
• Context is the failure point. The liver-test summaries were criticized for leaving out the qualifications a careful page would include. Pages that state who a figure applies to, and when it does not, are safer sources to cite.
• The sources behind an answer decide its quality. Google defended its summaries by pointing to the high-quality websites behind them. In YMYL categories, being one of those websites is the goal.

What gets cited in YMYL answers

On high-stakes questions, AI answers tend to draw on sources whose authority is easy to verify: regulators and government agencies, hospitals and clinical bodies, nonprofit and industry associations, and established publishers with named, credentialed authors. Brands appear too, most often where the question is about choosing a product or provider, and where the brand's own pages answer the question carefully and with evidence.

The practical consequence for a fintech, insurer, health or legal brand is that you compete with institutions on trust as well as with competitors on relevance. That is a harder bar, and also a clear one. Our analysis of why AI engines choose some brands and not others covers the general pattern; in YMYL categories, the trust signals below carry more weight than anywhere else.

How YMYL brands earn AI visibility

1. Name the experts behind your content. Show who wrote or reviewed each page, with credentials that match the topic: a licensed advisor for financial guidance, a clinician for health content, a qualified lawyer for legal explainers.
2. Cite primary sources. Link to the regulator, the study or the official dataset, and say what the figure applies to. Answers built from your page inherit its care.
3. Keep facts current. Prices, fees, rates, eligibility rules and product claims change. Outdated figures are the fastest way to make an AI answer about your brand wrong.
4. Publish clear regulatory disclosures. Licensing, supervision and risk statements are trust signals for readers, raters and the systems that summarize your pages.
5. Keep your company facts identical across the web. Your site, profiles, directories and structured data should agree. Structured data helps here; see our guide to structured data for AI search.
6. Earn mentions on the sites AI already trusts. Coverage in respected trade publications, inclusion in independent comparisons and reviews on recognized platforms all add the outside validation that YMYL topics demand.

For the page-level tactics that apply to Google's AI features in general, see our guide to earning visibility in AI Overviews.

What to avoid on YMYL topics

Thin or unreviewed AI-written content on high-stakes topics is the biggest risk. Google's guidelines warn against content produced at scale without care, and an AI engine has every reason to prefer a careful institutional page over a fast, generic one. Avoid unverified claims, figures without a date or source, and company details that differ between your site and your profiles.

Measuring visibility in a YMYL category

In regulated categories, being mentioned is not enough to count as a win. Track whether AI answers recommend you, mention you neutrally or advise against you, and treat disclaimers and declined answers as a separate signal: an answer that refuses to recommend any provider is a different situation from one that recommends a competitor. Our guide on how to monitor AI search visibility sets out the full framework, and AI share of voice shows how to compare yourself with competitors.

Tracking these answers prompt by prompt, across engines and markets, is the work Omnia does every day, so YMYL brands can see whether they are trusted, ignored or warned against, and act on it.

Start with your free 14-day trial with Omnia.