A summary judgment brief filed September 17 in the publishers’ copyright case against OpenAI and Microsoft puts AI link clicks at the center of the legal dispute. Microsoft data cited in the filing shows click-through rates (CTR, the share of times a URL gets clicked relative to how often it appears) were 87% to 93% lower in Bing Chat than in Bing web search for The New York Times’s sites. Internal messages from OpenAI suggest the company understood this dynamic as early as February 2023.
What Did OpenAI’s Own Engineers Say About AI Link Clicks?
The publishers’ combined brief quotes an OpenAI software engineer writing in February 2023: “no matter how prominently we show the links, users won’t click.” The same filing quotes Nick Turley, OpenAI’s head of ChatGPT, saying there was “no good reason to click” on a link to the original source once ChatGPT gave an answer. Both quotes were reported by Search Engine Journal based on the court document.
These are one side’s arguments. The federal court in New York has not yet ruled. Judge Sidney H. Stein’s September 24 order set October 23 for opposition briefs, October 30 for amicus briefs, and November 20 for replies.
OpenAI and Microsoft filed their own summary judgment motions on September 4, arguing their use of news content is fair use. Per Axios, OpenAI’s motion cited studies suggesting ChatGPT outputs don’t create commercial substitutes for The Times’s work. A Microsoft spokesperson told TheWrap the company’s filings explain “why Copilot is not a substitute for publishers’ journalism.” OpenAI did not respond to TheWrap’s request for comment.
What Does the Microsoft CTR Data Show?
The publishers’ brief cites Microsoft data comparing CTR in Bing Chat versus Bing web search for three publisher groups. All three show substantial drops in the chat product, per Search Engine Journal’s reporting on the filing:
- The New York Times’s websites: CTR 87% to 93% lower in Bing Chat
- Daily News plaintiffs’ sites: 83% to 91% lower
- Ziff Davis sites: 51% to 94% lower
Several caveats apply. The passage doesn’t say when Microsoft collected this data, how many queries or impressions it covered, or what the low and high ends of each range represent. Microsoft renamed Bing Chat to Copilot in November 2023, but that product name change alone doesn’t date the data. The ranges measure CTR, not absolute traffic volumes or visits.
The publishers argue these reductions “clearly show substitution” and state that “common sense suffices” to establish harm without additional data. OpenAI and Microsoft dispute the substitution framing. Both sides’ arguments are now in the court record.
Why Do AI Visibility Tools Show Impressions but Not Clicks?
The measurement gap runs across both major platforms. Microsoft’s AI Performance report in Bing Webmaster Tools tracks citations (how often a URL appears in an AI-generated answer), not clicks. Microsoft’s own documentation states citation metrics differ from traditional metrics that measure clicks and traffic. When Microsoft added Citation Share, it noted explicitly the metric doesn’t represent traffic share.
Google followed a similar path: the company launched new Search Generative AI performance reports in Search Console, covering AI Overviews, AI Mode, and Discover. Per Google’s announcement, these reports track impressions, along with Pages, Countries, Devices, and Dates. The reports are rolling out to a subset of websites. Search Engine Journal notes the report doesn’t list clicks.
The pattern is consistent: site owners can now measure how often their content surfaces in AI answers on both Google and Microsoft’s platforms. They cannot, through these standard tools, see how many users clicked through from those answers.
What This Means for AI-Search Visibility
The plain version: AI answers are designed to be complete, and when users get what they need inside the chat window, the link becomes optional. That is the structural reality the court data illustrates.
A concrete example: if someone asks an AI chatbot to compare two software tools and the AI gives a detailed breakdown with a recommendation, most users will act on that answer without visiting either product’s site. The AI has resolved the question. The link is still there, it’s just no longer necessary to reach the conclusion. The OpenAI engineer’s February 2023 message suggests this was understood internally before AI search products went mainstream.
This creates a split that the current measurement tools have quietly confirmed. Both Google and Microsoft have launched AI-specific visibility metrics that track appearances in AI answers, separate from click metrics. When Microsoft added Citation Share, it said upfront it doesn’t measure traffic. Google’s generative AI report shows impressions and, as Search Engine Journal notes, doesn’t list clicks. In both cases the metric counts appearances rather than visits, because citations and clicks are genuinely different outcomes in AI-assisted search.
One important nuance the sources don’t address: the Microsoft CTR data in the filing covers journalistic content specifically, newspapers and news publishers. It doesn’t tell us whether the same CTR gap applies to product pages, how-to content, or brand coverage. Publishers produce content that is often self-contained once summarized; other content types (product listings, booking pages, contact information) may behave differently because the AI answer can’t substitute for the action the user needs to take on-site. The court record currently gives us data for one content category.
Hingewise’s read: treating citation share as a proxy for traffic is a measurement error, not just a conservative assumption. The more useful question is which page types and query intents still drive clicks even when an AI answer is present, and whether your content strategy is weighted toward those. That’s where the measurement gap actually matters for brands.
What to Check Before Treating AI Impressions as Traffic
- Confirm whether your Search Console account has access to the new generative AI performance reports (rolling out to a subset of sites).
- Check whether your Bing Webmaster Tools data separates chat and web clicks, Microsoft combined them in 2023 before introducing separate citation metrics.
- Compare your citation share trend against actual referral traffic from AI sources over the same period to see whether the two move together or diverge.
- Identify which page types appear in AI answers most often, informational articles, product pages, and brand pages may have very different click conversion rates.
- In any reporting, note the gap between “impressions in AI features” and “clicks from AI features” explicitly; do not use one as a substitute for the other.
- When the October 23 opposition filings are published, check whether OpenAI or Microsoft add context on how the CTR data was collected or what the ranges represent.
The next public milestone in the case is October 23, when OpenAI and Microsoft submit their oppositions. Whether those filings address the engineer quotes, Turley’s comment, or the CTR methodology will add substantially to the public record on how AI-generated answers handle outbound links.
Sources: Search Engine Journal (reporting on court filing, September 2026); Google Search Central Blog (June 2026). Author: Lam Nguyen, Hingewise.
Sources
This article was drafted with AI assistance from the sources cited above and checked against them before publishing. The "What this means for AI-search visibility" section is Hingewise's editorial analysis, built on the frameworks and positions we use in our GEO work. How we work: Editorial policy.