Court Filing Quotes OpenAI Engineer Saying Users "Won't Click" Source Links

A court filing in a landmark copyright case quotes an OpenAI engineer admitting users "won't click" source links — and Microsoft data suggests click-through rates in AI chat are dramatically lower than in traditional search.

The revelation emerged from a publishers' combined summary judgment brief filed September 17 in federal court in New York. The case pits major news organisations against OpenAI and Microsoft in one of the most consequential copyright battles in the history of digital media.

---

What the Court Documents Reveal

The brief quotes an unnamed OpenAI software engineer who wrote in February 2023 that "no matter how prominently we show the links, users won't click." That admission carries significant weight in a case built around whether AI chatbots are substituting for — rather than directing traffic to — original journalism.

The filing goes further. It also quotes Nick Turley, OpenAI's head of ChatGPT product, who reportedly said there was "no good reason to click" on a link to the original source once ChatGPT had already delivered an answer. Together, the two statements form a core pillar of the publishers' argument that AI-generated responses actively displace web traffic rather than supplement it.

These internal admissions are particularly damaging because they suggest the substitution effect was understood — and arguably accepted — within these organisations long before it became the subject of federal litigation. Unlike external critics speculating about AI's impact on publishing, these are the voices of people who built and managed the products in question.

The New York Times is among the plaintiffs in the consolidated case before Judge Sidney H. Stein. Other plaintiffs include the Daily News and Ziff Davis publications. The filing represents one side's legal argument and the court has not yet ruled on the matter.

---

The Click-Through Rate Data at the Centre of the Dispute

The publishers' brief cites Microsoft data comparing click-through rates for plaintiff websites in Bing Chat versus Bing Web Search. The numbers are striking.

The CTR Figures

For The New York Times's websites, the filing reports CTR was 87% to 93% lower in Bing Chat than in Bing Web Search. For Daily News plaintiffs' sites, the drop ranged from 83% to 91%. For Ziff Davis properties, the range was 51% to 94% lower.

The filing defines CTR as the percentage of clicks to a URL relative to its impressions — meaning how often a link appeared in a response or search results page compared to how often users actually clicked it. Publishers argue these reductions "clearly show substitution" and contend that "common sense suffices" even without proving harm through data alone.

For publishers and content strategists already working to increase organic traffic to their websites, these figures represent more than a legal data point — they quantify a structural shift in how audiences access information online.

The Gaps in the Data

However, significant gaps exist in the data as presented. The filing does not specify:

• When Microsoft collected the figures
• How many queries or impressions were involved
• How the data was selected
• What accounts for the low and high ends of each percentage range

Microsoft renamed Bing Chat to Copilot in November 2023, but nothing in the filing ties the figures to the current Copilot product.

Citations Are Not Clicks

A critical distinction shapes how this data should be interpreted. Being cited in an AI response and being clicked are two entirely different measurements. Microsoft's AI Performance report in Bing Webmaster Tools counts citations, and the company has stated that citation metrics differ from traditional search metrics measuring rankings, clicks and traffic. When Microsoft introduced Citation Share in June, it explicitly noted the metric does not represent traffic share.

This distinction matters enormously for anyone drawing conclusions from platform-level reporting. A high citation rate in an AI product can coexist with near-zero referral traffic — and the Microsoft data cited in the filing suggests that is precisely what is happening.

---

What This Means for Publishers and the Broader AI Debate

OpenAI and Microsoft Push Back

OpenAI and Microsoft filed their own summary judgment motions on September 4 — weeks before the publishers' brief landed. Both companies argue their use of news content constitutes fair use under copyright law.

According to Axios reporting on OpenAI's motion, the company cited studies suggesting ChatGPT outputs do not create commercial substitutes for The Times's work. Microsoft made similar claims about Copilot. A Microsoft spokesperson told TheWrap that the company's filings explain "why Copilot is not a substitute for publishers' journalism." OpenAI did not immediately respond to a request for comment from TheWrap.

Watch for whether OpenAI and Microsoft address the engineer quote, Turley's comment, or the CTR data directly when their opposition briefs are due October 23. Amicus briefs are scheduled for October 30 and reply briefs for November 20.

The Existential Pressure on Digital Publishing

The case is unfolding as AI search tools have become mainstream — a dynamic the publishing industry compares to a slow-moving existential threat. If AI systems deliver complete answers without prompting users to visit original sources, the economic model underpinning digital journalism faces serious pressure. This is the media industry's own Napster moment — except the content being consumed without compensation is news itself, produced at significant cost by professional journalists.

The underlying tension is not simply legal — it is economic. Publishers invest in reporting, editorial oversight and distribution infrastructure. If AI products extract value from that investment while eliminating the click that funds it, the revenue model collapses regardless of how copyright law is ultimately interpreted. Understanding the risks and challenges AI poses to established business models is increasingly essential for any organisation operating in the digital content space.

Practical Steps for Publishers and Marketers

For anyone managing a digital presence, the gap between AI citations and actual clicks is no longer theoretical. It is now the subject of federal litigation backed by internal data from the companies building these tools.

Publishers and content marketers can act on this information now:

1. Track AI referral traffic separately from direct traffic using source-tagging methods, since platforms like ChatGPT, Claude and Perplexity currently funnel visits into "direct/other" buckets, obscuring the real picture.
2. Monitor citation share metrics in Bing Webmaster Tools while understanding they do not equal traffic — and should not be treated as a proxy for audience reach.
3. Pay close attention to the October 23 opposition filings from OpenAI and Microsoft, as they may contain additional data or context that either supports or complicates the CTR figures currently driving headlines.

The way AI is reshaping customer-facing discovery also has implications beyond publishing. Businesses deploying AI tools in customer service and engagement should consider how the same dynamic — complete answers replacing the click — affects their own funnels and conversion pathways.

What Comes Next

The outcome of this case could reshape how AI companies are required to credit and compensate original publishers. For a deeper examination of the legal arguments, the Reuters legal affairs desk has been tracking the case closely as it develops.

This is one of the most consequential legal proceedings for the future of digital media and search — and the internal quotes now on the public record suggest the companies at the centre of it understood the stakes long before a judge was asked to weigh in.