The participation of ChatGPT in the volume of queries to artificial intelligence assistants fell from 70% to 50% between January and June 2026, while Gemini rose from 17% to 30% and Claude from 2% to 11%, according to the AI Intelligence report for the second quarter from Comscore, published by Digiday on September 22 in a note by Sara Guaglione.
This movement does not mean that the use of AI assistants is decreasing. The absolute figures confirm that this is a redistribution of the market and not a contraction.
In June 2026, ChatGPT recorded 168 million desktop conversations, Gemini 89 million, and Claude 22.3 million, according to the same report.
User visitation also grew on both devices: 35% on desktop compared to 25% in January 2025, and 29% on mobile compared to 12% at the same point of comparison.
The change also affects the search engine. The presence of AI-generated views within Google search results rose from 25.8% in July 2025 to 39.4% in June 2026.
The Competition is No Longer Just Among Assistants
Smriti Sharma, director of analytics and technology at Comscore, provided insight on the movement from a consumer behavior perspective:
The message is not that ChatGPT is winning the race. The message is that artificial intelligence assistants are collectively becoming a layer of consumer behavior, and consumers are developing distinct relationships with different assistants.
The implication for brands, publishers, and platforms is significant: the consumer is not necessarily developing an exclusive relationship with a single assistant. They are incorporating different tools at various moments of search, discovery, and decision-making.
Being a Source Does Not Mean Receiving the Click
The background that changes the reading is the asymmetry between being a source and being cited.
The report documents that TripAdvisor appears as the source of information in 61% of relevant cases, but receives visible attribution in only 21%.
The gap is 40 percentage points.
This differential describes one of the economic problems faced by the content ecosystem in the era of assisted discovery: content can sustain a response without necessarily receiving the visit.
For publishers and brands, this modifies an equation that for years was based on a relatively direct relationship between content, search, click, and traffic.
For Latin America, Fragmentation Has an Operational Consequence
For media and brands in Latin America, the fragmentation data has a concrete consequence: optimizing for a single assistant becomes unviable when none concentrates more than half of the volume.
Moreover, each platform builds its own index and its own citation criteria.
This forces an expansion of the distribution logic. It is no longer just about positioning content in a search engine or optimizing it for a specific platform, but about building sufficient presence and authority to appear in different layers of the new discovery ecosystem.
The Data That Marketing Should Note
From the editorial perspective of next+, the figure that a marketing director should note is not ChatGPT's decline, but the 39.4% of Google results with AI-generated responses.
The difference is strategic.
The fragmentation among assistants is a distribution problem that is solved with multiple presence.
In contrast, the response generated within the search engine is a business model problem, because it occurs precisely in the channel that historically delivered the click.
Four out of ten searches already return an answer before the link.
And that transformation has a direct consequence for content strategies: no strategy compensates for the potential loss of traffic simply by producing more content.
The challenge is to build content with enough authority, relevance, structure, and differential value to compete not only for the user's attention but also to be part of the answers that new systems use to solve their searches.