Semrush: ChatGPT Has Raised Its Share of News Sources Seven Months in a Row

Semrush reported on Tuesday, October 13, that ChatGPT relies on news sites more heavily than any other major AI platform, in both the US and the UK. From January to July, the share of news among the sources OpenAI's assistant cites in its answers increased every month, seven periods in a row. Gemini, along with Google's AI Mode and AI Overviews, fell behind in this comparison. In the British segment, the tilt toward news is more pronounced than in the American one.

This is not a story about a single chart. It shows what AI assistants build their answers from and who gets the attention when users no longer open ten blue links.

What the report found

Semrush compared the mix of cited sources across several platforms. From the available description of the study, the picture looks like this:

• observation period: January–July, seven months;
• markets: the US and the UK;
• platforms: ChatGPT, Gemini, AI Mode, AI Overviews;
• leader by share of news sources: ChatGPT in both markets;
• ChatGPT trend: the share of news grew in every month of the period;
• regional difference: the pull toward news is stronger in the UK.

The available excerpt of the publication contains no exact percentages, so the size of the gap between platforms cannot yet be compared in numbers. The direction and consistency of the trend are known: growth came with no gaps.

Where the skew comes from

In the part of the material that is available, Semrush does not explain the causes, so what follows is only hypotheses.

The first concerns licensing. OpenAI has signed agreements with a number of major publishers, and it is logical that such partners appear more often in answers with attribution. The second hypothesis concerns how the products themselves are built. AI Overviews and AI Mode run on top of Google's search index, where results include many forums, videos, commercial pages and reference resources. News competes there with a much more varied set of formats. ChatGPT, judging by the data, more often picks journalistic material when answering queries.

The third hypothesis explains the British skew. The country's media market is concentrated around a few strong brands, and it is easier for models to find and rank recognizable sources. This is an assumption, not a finding of the study.

There is also a caveat. Share of citations is not the same as traffic. A link in an assistant's answer does not guarantee that the user will click it. For publishers, this is a metric of presence, not of revenue.

What publishers and marketers should do

First: measure visibility by platform separately. An average across AI search tells you nothing if ChatGPT and Google's products draw their sources from different baskets. The Semrush report confirms this with data from two markets.

Second: check bot access. If OpenAI's crawlers, including OAI-SearchBot, are blocked in robots.txt, a piece of content has a lower chance of being cited. The decision on access should be made deliberately and together with legal counsel, especially if licensing talks are under way.

Third: brands that are not media outlets should treat news mentions as a channel into ChatGPT's answers. If the assistant prefers to take data from journalistic material, then publications in trade press influence how the model describes a company. This is an argument in favor of digital PR.

Fourth: track the trend over time, not a single measurement. Seven consecutive months of growth point to a trend, but the source shares of AI platforms shift as models and search systems are updated. Monthly monitoring of your own key queries in the US, the UK and other important markets is the most useful approach.