Muck Rack：2026 AI “阅读习惯”洞察报告：基于对超过2500万条 AI 引用链接的系统性分析（英文版）（43页）.pdf

1、What is AI Reading?May 2026Journalism and earned media remain the foundation of AI citations:About 99%of links cited by AI come from non-paid sources.Journalism accounts for about 27%of all citations.AI providers have distinct citation habits:ChatGPT cites sources in 96%of responses,averaging 5 cita

2、tions per response.On the other hand,Claude cites more selectively in only 55%of responses,but averages 13 citations when it does.And Gemini cites sources in 82%of responses with an average of 8 citations.ChatGPTs number one cited domain is Wikipedia,Claudes is PubMed Central,a free database of scie

3、ntific research,and Geminis is Reddit.What you ask determines what AI cites:Questions about industry trends drive journalism citations at more than twice the rate of how-to queries.Press releases appear almost exclusively in industry trend responses too.Journalism citations are distributed across th

4、ousands of outlets,with one exception:Axios appears in ChatGPTs top three cited domains across 13 of 17 industries.The Associated Press appears in the top three for one industry while The New York Times and Reuters do not appear in the top three most cited sources for any industry.Executive SummaryM

5、ethodology This study examines how generative AI systems cite sources when answering realistic consumer prompts.We submitted prompts to three AI providers:ChatGPT(OpenAI),Claude(Anthropic),and Gemini(Google).Sometimes prompts specifically mention companies by name,sometimes they do not.We recorded e

6、very URL each model cited.To accomplish this,we constructed a large,diverse prompt set and executed it across several web-enabled language models,followed by systematic analysis of the responses and the cited links.The prompts span a variety of industries and subject matter.Generative AI systems are