Claims that OpenAI users will not click links run ahead of the record. A court filing can surface important evidence, but it is still a party’s argument, not a broad benchmark of user behavior. The safer question is narrower: whether AI chat often reduces clicks in certain settings compared with classic search.
Table of Contents
That distinction matters because traffic risk, trust, and discovery do not move in lockstep. For publishers and marketers, the real issue is scope, not slogan.
What the court filing actually says about AI links and user clicks
The safest reading is narrow: this does not prove that AI users will not click links. A court filing may present a party’s argument, selected exhibits, and limited examples. Legal filings are not broad benchmarks of user behavior.
Even if a filing points to lower clicking in an AI interface, it describes one setting, time, and measurement choice. It does not show that users never visit sources. Nor does it show that every AI answer replaces a search.
The practical takeaway is restraint. Treat the phrase OpenAI wont click links as a disputed summary, not a settled finding. The stronger question is what behavior was measured, in which product, and against what comparison point.
That distinction keeps the claim tied to its actual evidence and scope.
How strong is the evidence that chat users click fewer links than search users
Evidence for fewer clicks in chat is directional, not final. The safest reading is weaker than the headline claim.
- One reason is the comparison problem. Chat, AI summaries, and classic search do not always serve the same task. Lower clicking may reflect a different interface, not a different user preference.
- UCLA Anderson Review points to research on more than 1 million U.S. desktop users in 2022 and 2023. It also reports that traditional-result clicks fell to 8% when an AI summary appeared. That supports a lower-click pattern, but only within a defined setting and period.
- The practical conclusion is modest. There is enough evidence to take OpenAI wont click links seriously as a traffic risk. However, it is not enough to treat the phrase as a universal rule across all users, devices, and query types.
Why clicks drop when an AI answer feels complete on the page
Completion on the page changes the click decision because the visit feels unnecessary for every query.
- When an answer resolves a simple fact, the next step disappears. A click usually happens when the page leaves a meaningful gap, not when it feels complete.
- Format also matters. Chat-style replies combine a summary, context, and space for follow-up. Users can keep working in one interface instead of opening another tab.
- This effect reflects task compression, not proof of stronger loyalty. Fewer visits may result from less friction on basic questions, while harder tasks still give users reasons to leave.
- For publishers and marketers, the main risk is lost traffic on answerable queries. Pages that add depth, proof, tools, or original reporting still give users a clear reason to click.
Does lower click-through mean users trust AI more, or just need fewer visits
Lower click-through does not automatically signal deeper trust. It can also mean the interface solved enough of the task before a visit felt necessary.
- Trust and reduced visiting measure different behaviors. A user may accept a quick answer for one step, then still doubt the system on higher-stakes decisions.
- Research posted on arXiv about human-AI collaborative annotation suggests explanations can influence human judgment. That matters here because agreement with an answer can reflect persuasive presentation, not a broad decision to trust the tool itself.
- For marketers and publishers, the practical read is restraint. A lower click rate is better treated as a sign of task completion or convenience unless other signals show sustained confidence, repeat use, or reliance in more demanding contexts.
Where the claim may not hold up across tasks, devices, and query types
Context matters because clicking changes when the task, risk, and answer quality change. These boundaries limit any universal no-click rule.
- Simple prompts may end in chat, but not every query is simple. Device context matters, and so does query type.
- Current or changing topics can push users outward. The investigation noted that models trained only through September 2023 or October 2023 sometimes warned that newer developments might exist.
- High-stakes searches also resist the no-click claim. The Office of the Privacy Commissioner of Canada found examples of inaccurate and harmful statements, so some tasks still create reasons to verify elsewhere.
- Interface details matter too. The same investigation found the accuracy disclaimer at the bottom in small gray text. Device layout and notice visibility may affect whether a user clicks for confirmation.
What marketers can still measure when traffic shifts from clicks to assisted actions
Marketers can still track outcome signals that show whether AI-assisted journeys create business value, even when visits decline.
- assisted conversions: Watch conversions that follow direct visits, branded search, or return sessions. Click loss matters less when qualified actions hold steady or rise.
- Engaged demand: Measure demo requests, trial starts, email sign-ups, and sales-ready lead quality. These signals show whether visibility is producing real interest instead of empty impressions.
- brand lift proxies: Track branded queries, direct traffic, and repeat visitors together. None proves causation alone, but movement across all three is more useful.
- Content usefulness: Monitor scroll depth, tool use, downloads, and page-to-page progression. If audiences continue deeper, the page is likely adding value beyond summary answers for the same task.
How brand discovery changes when answers summarize sources instead of sending visits
When an answer gives the gist first, discovery shifts from referral to recall. The audience may learn a name, claim, or category without opening the source page. That changes what visibility means. A brand can enter the decision path yet remain absent from session-based analytics.
The tradeoff is clear. Summary answers may widen exposure near the journey’s top, but they reduce chances to show depth, proof, and on-site distinction. In that setting, memorable framing matters more than raw impressions alone.
Distinct language, clear expertise, and reusable facts may survive the summary layer more often. The practical move is to view discovery as a two-step outcome: first, being surfaced; then, being remembered through branded search, direct visits, or return demand.
Taken together, the claim is too broad. AI answers can reduce clicks in some settings, especially on simple queries that feel complete on the page. But that pattern does not prove users will not click links, trust AI more, or behave the same across devices and tasks.
The practical reading is narrower: traffic risk is real, yet it varies with query type, depth, and verification needs. The strongest response is to value measures beyond raw visits and create pages that offer depth, proof, tools, or original reporting.
