Data study · Patents behind AI search
12.9% of Generative AI Search Patent Families Describe Using a Person's History, Profile or Preferences
292 of 2,269 AI search patent families describe personalization. User profiles and preferences lead, and conversation history appears in 47 families.
Analysis of 2,269 generative AI search patent families, 292 describing personalization · Google Patents Public Data, October 2026
Key findings
- 1
292 of 2,269 generative AI search patent families (12.9%) describe personalizing results with a user's history, profile or preferences.
- 2
The share rose from 10.6% of families published in 2024 (42 of 395) to 13.4% in 2026 so far (106 of 794).
- 3
User profiles appear in 51 of the 292 (17.5%) and stated preferences in 50 (17.1%).
- 4
Conversation or chat history appears in 47 (16.1%), more than browsing history (14) and purchase history (8) combined.
- 5
222 of the 292 (76.0%) include a US publication, compared with 1,253 of all 2,269 AI search families (55.2%).
- 6
Google (25) and Microsoft (21) hold the most personalization families.
Summary
Picture two people in the same town asking an AI assistant to recommend a roofer and getting different names. One of them has been chatting with the assistant about storm damage for a week. The other just moved in. Personalization can be one reason answers differ.
We checked which generative AI search patent families describe using personal signals. 292 of 2,269 do, or 12.9%. The share has crept up each year, from 10.6% of 2024 publications to 13.4% in 2026 so far.
Profiles, preferences and past conversations are the signals named most. A patent shows what was filed, not what is live in any product.
What we measured
We pulled every publication in Google Patents Public Data (patents-public-data.patents.publications) published from January 1, 2018 to September 17, 2026 that carries at least one CPC code in G06F16 (information retrieval), G06F40 (natural language processing) or G06N (AI and machine learning) and has an English abstract. That is 2,170,848 publications in 1,214,491 patent families. For each publication we kept the English title, the English abstract and up to the first 20,000 characters of the English claims. English claims exist for 488,406 publications, mostly US, European and international (WO) filings, so text matching is deeper for those offices than for Chinese, Korean or Japanese filings, which we matched on title and abstract.
Within the 2,269 families that describe both generative AI and search, we flagged personalization when the text mentions personalizing, a user profile, user preferences, or search, browsing, purchase, interaction, conversation, chat, dialog or query history, or past or previous queries, searches, purchases or conversations. We then counted specific signals separately. A family can name several signals, and some match only general personalization wording.
About one in eight AI search patents describe personalization
We counted personalization families by first publication year.
Share (%)
| Year | Personalization families | All AI search families | Share (%) |
|---|---|---|---|
| 2,023 | 7 | 48 | 14.6 |
| 2,024 | 42 | 395 | 10.6 |
| 2,025 | 131 | 1,005 | 13 |
| 2,026 | 106 | 794 | 13.4 |
Source: aisearchoptimization.net analysis of 2,269 generative AI search patent families, 292 describing personalization · Google Patents Public Data, October 2026
Counts before 2023 are tiny (6 families in 2021 and 2022 combined). By first filing year, 78 were first filed in 2023, 139 in 2024 and 57 in 2025 so far.
Which personal signals the patents name
We looked for specific signals inside the 292 personalization families.
Families
| Signal | Families | Share of 292 (%) |
|---|---|---|
| User profile | 51 | 17.5 |
| User preferences | 50 | 17.1 |
| Conversation or chat history | 47 | 16.1 |
| Search or query history | 44 | 15.1 |
| Browsing or click history | 14 | 4.8 |
| User location | 9 | 3.1 |
| Purchase history | 8 | 2.7 |
Source: aisearchoptimization.net analysis of 2,269 generative AI search patent families, 292 describing personalization · Google Patents Public Data, October 2026
Conversation history is a signal that only exists in chat-style search. It appears in 47 families, while purchase history appears in 8.
Personalization filings lean toward US publications
We compared publishing offices for personalization families with the full AI search set.
| Measure | Personalization families | All AI search families |
|---|---|---|
| Include a US publication | 222 of 292 (76.0%) | 1,253 of 2,269 (55.2%) |
| Include a Chinese publication | 91 of 292 (31.2%) | 1,049 of 2,269 (46.2%) |
| 25 | ||
| Microsoft | 21 | |
| Dell | 9 | |
| eBay | 6 | |
| Intuit | 6 | |
| Maplebear (Instacart) | 6 |
Source: aisearchoptimization.net analysis of 2,269 generative AI search patent families, 292 describing personalization · Google Patents Public Data, October 2026
Part of the US lean comes from our method, because US filings include English claims where signals like browsing history are spelled out.
What this means
Personalized answers mean no single test query tells you how visible you are. What one person sees depends partly on what they asked before. Checking many prompts phrased the way real customers talk gives a truer picture than checking one. That is how AI search optimization results should be measured too.
Make your business easy to match to specific needs: list the jobs you do, the areas you serve and who you serve best, so a system can connect you to a person's situation. A free AI visibility audit can show how you appear across a range of customer prompts, and any AI SEO agency should test that range, not a single query.
Limits
A patent shows what was filed, not what is live in any product.
Patents usually publish about 18 months after the first filing, so families first filed in 2025 and 2026 are heavily undercounted, and 2026 publication counts cover January 1 to September 17, 2026 only.
Signals are matched by keywords and may describe options listed in a patent rather than core methods.
Claims-based matching is deeper for US, European and WO filings.
Owner counts use harmonized assignee names; individual owners are not named.
Embed this research
Add the charts from this study to your page. The embed links back here.
<iframe src="https://aisearchoptimization.net/embed/data-studies/ai-search-personalization-patents" title="12.9% of AI Search Patents Use Personal History" width="100%" height="720" style="border:0" loading="lazy"></iframe>
Cite this study
APA
Grenyo, K. (2026). 12.9% of AI Search Patents Use Personal History. AI Search Optimization. https://aisearchoptimization.net/resources/data-studies/ai-search-personalization-patents
Plain link
12.9% of AI Search Patents Use Personal History - AI Search Optimization - https://aisearchoptimization.net/resources/data-studies/ai-search-personalization-patents
Frequently asked questions
How many AI search patents describe personalization?
292 of 2,269 generative AI search patent families, or 12.9%. In 2026 so far it is 106 of 794 (13.4%).
Which personal signals come up most?
User profiles (51 families), user preferences (50), conversation history (47) and search history (44). Purchase history appears in only 8.
Does this mean AI assistants track my search history?
Not necessarily. A patent shows what was filed, not what is live in any product.
Why do AI answers differ between people?
Many reasons are possible, and personalization is one approach described in these filings. Wording, location and earlier conversation can all change what a system returns.
Related studies
- Patents behind AI searchFreshness Shows Up in Only 3.6% of AI Ranking Patents
- Patents behind AI searchQuery Fan-Out Patents: 44.1% Now Use Generative AI
- Patents behind AI searchSearch Patents 2026: 39.9% Now Mention Generative AI
- Patents behind AI searchOnly 3.9% of AI Search Patents Describe Citing Sources
See where AI search leaves you out
Run the free AI visibility audit. If you want the fixes done for you, the plan is a flat $299/month.