When Frank Pine recently searched Google for a news article, he found AI-generated summaries at the top of his results, pushing the actual links further down. This experience, which initially annoyed Pine, executive editor of Media News Group and Tribune Publishing, has now become a significant concern.

Google announced in May that these AI-generated summaries, called AI Overviews, would be available to all users in the United States. These summaries compile content from various news sites and blogs on the searched topic, and media executives fear this could drastically reduce traffic to their sites from Google.

“It potentially chokes off the original creators of the content,” Pine commented, describing AI Overviews as another step towards generative AI replacing the publications they depend on.

Media companies now face a dilemma. They rely heavily on being listed in Google’s search results, which can drive over half of their traffic. However, participating in these listings means Google can use their content in AI Overviews. If publishers block Google’s web crawler from using their content, their links will appear without descriptions, making them less appealing to click. Opting out of Google’s index entirely could be devastating for their business.

Renn Turiano, head of product at Gannett, the largest newspaper publisher in the US, noted that while refusing Google’s indexing is not viable, AI Overviews are “greatly detrimental to everyone apart from Google, but especially to consumers, smaller publishers, and businesses that use search results.”

Google maintains that its search engine continues to drive billions of visits to websites, adding value to publishers. Liz Reid, Google’s vice president of search, noted that during testing, AI Overviews often led to more clicks on the links they featured compared to traditional blue links.

However, following some high-profile errors, such as suggesting users eat rocks for a balanced diet, Google announced it would limit AI Overviews to fewer search results while continuing to refine the system.

The introduction of AI Overviews has exacerbated tensions between tech companies and publishers, who are already engaged in legal battles over the use of news content to train AI models. Media News Group and Tribune Publishing, alongside The New York Times, have sued OpenAI and Microsoft for copyright infringement.

Publishers are now emphasising the need to develop direct relationships with readers, focusing on digital subscriptions and direct site visits to reduce reliance on search engines. Nicholas Thompson, CEO of The Atlantic, mentioned investing more in direct reader engagement through channels like email newsletters.

The debate over fair compensation for content used to train AI models continues, with some publishers securing deals with AI firms. OpenAI, for instance, has agreed to pay for content from several major publishers. In contrast, Google has yet to make similar agreements, arguing that paying for content would undermine the open web.

Roger Lynch, CEO of Condé Nast, stressed that while embracing the future of AI is inevitable, it must happen on terms that protect content creators. He suggested that legislative action might be necessary to ensure fair compensation for the use of content in AI training.

Despite concerns, some publishers see participation in AI Overviews as crucial. Thompson of The Atlantic noted that being part of Google’s summaries might help mitigate the expected decline in traffic as Google implements these changes.

As the industry navigates these challenges, media executives continue to seek solutions that balance innovation with the sustainability of their business models.