One-tenth of a per cent recognition

Google is now paying some publishers for content featured in AI-generated responses. That sounds like a breakthrough. On closer inspection, however, it becomes clear just how unevenly the bargaining power is distributed.

There are figures you have to read twice. “For several small and medium-sized websites in Google’s new pilot programme, the payments amount to less than one-tenth of a per cent of their advertising revenue.” One-tenth of a per cent. For most publishers, that’s not worth taking seriously.

What’s it all about? Google is currently testing what is known as the ‘AI Contribution Pilot’. According to media reports, around 100 publishers are taking part. Those involved receive an additional widget in Search Console that shows their monthly AI revenue. Payments are made when a publisher’s own content contributes ‘significantly’ to answers in Google’s AI products, including AI Overviews, AI Mode and Gemini. The reported amounts vary considerably. Some smaller websites have received less than $1,000 over several months, whilst individual larger participants are said to be earning more than a million dollars a year. According to the information available so far, another participant has earned between $50,000 and $60,000.

At first glance, this sounds like a new business model for publishers. And indeed, it is remarkable that Google is even beginning to pay directly for the use of journalistic content in its AI responses. For years, the relationship between Google and publishers was based primarily on a different principle: Google indexes content, displays it in search results and redirects users to the websites. In return, the publisher gains reach and potential advertising revenue. With AI-generated answers, this relationship is changing. If an answer summarises the key information directly on the Google page, the user may not need to visit a publisher’s website.

The fact that Google is now paying for the use of content is therefore, in principle, an exciting development. In doing so, the company is at least acknowledging that content has economic value when it is directly incorporated into AI-generated responses. For publishers, this sets an important precedent. The amount of the current payments is another matter entirely.

The pilot programme could therefore be seen as a step forward. And to a certain extent, it is. Publishers have long been waiting for the use of their content by AI systems to generate a transparent financial return. From the perspective of some participants, direct payments are also more attractive than hoping for a return to the referral traffic of the past. If, in future, Google were to pass on a share of the revenue from AI products to the sources of the information used, that would at least be a step towards a new form of monetisation.

However, it is currently difficult to see what criteria are used to determine this figure. This is where the problem with the pilot programme lies.

According to Google, only content that makes a ‘significant’ contribution to an AI-generated response counts. If content is subsequently used merely to confirm facts or as a link, this apparently does not give rise to any entitlement to remuneration. Google does not disclose exactly what constitutes a significant contribution or how this contribution is translated into a specific dollar amount.

This is difficult for publishers. It is impossible to plan, review or negotiate remuneration in a responsible manner when the logic behind its calculation is unknown. According to participants, the monthly amounts also fluctuate, in some cases considerably, without any clear explanation as to why. For a media company that needs to plan its content and investments in the long term, this represents a significant source of uncertainty.

The question of who actually determines the value here becomes even more interesting. A manager at one of the participating publishers describes the model as follows: Google uses the content whilst at the same time determining the value of that use. This means that a platform decides on the remuneration for content on which it itself depends for its own use.

From the publishers’ perspective, this is an unusual negotiating situation. In traditional licensing deals, the terms of use, scope, reach and price are agreed between two parties. In the Google pilot scheme, one party sets the rules of the programme, determines the metrics and uses these to calculate the payment. Publishers have little way of verifying whether this valuation corresponds to the actual economic value of the content.

There is also another important distinction to be made. Opting out of the remuneration scheme does not automatically prevent Google from using one’s own content for AI products. Participation in the pilot scheme and the fundamental control over how Google uses content are two separate issues. Google now offers technical options that allow publishers to restrict the use of their content for specific AI functions. However, how attractive this option is to an individual media organisation depends heavily on its reliance on Google Search and other Google products.

It is precisely this dependence that makes the discussion about remuneration so relevant. In theory, a small publisher can say no. In practice, however, it must take into account the consequences such a move might have for its visibility and reach. Large media companies are better placed to try to negotiate their own licensing agreements. Some publishers are said to have declined to take part in the pilot programme because they believe it will give them a stronger negotiating position for larger agreements.

In his well-known model, the economist Albert O. Hirschman described three possible responses when people or organisations are dissatisfied with an institution: exit, voice and loyalty. One can leave, speak out or stay.

Publishers face a similar dilemma. For many media companies, a complete withdrawal from Google’s ecosystem is economically unthinkable. It is possible to object, but this achieves little as long as the programme’s evaluation criteria remain opaque. And anyone who takes part must, for the time being, accept the terms set by Google.

That is precisely why the debate about Google’s pilot programme should not stop at the question of whether publishers will now receive payment for the use of AI. What matters is what rules emerge from this. A remuneration model can only become a viable business for publishers if it is clear how the value of content is determined and which forms of use are remunerated.

With this pilot programme, Google has taken a step that many publishers have been calling for for years. That should be acknowledged. At the same time, it would be too early to conclude that this represents a breakthrough for the monetisation of journalistic content. For many publishers, one-tenth of a per cent of advertising revenue is currently hardly worth mentioning. The far more significant question, however, is who will determine in future what our content is worth in an AI-generated response. And as long as this decision essentially lies with Google, the balance of power between the platform and publishers will remain unchanged.