The British regulator believes that the media must be able to refuse to be included in Google’s AI without suffering retaliation. A decision that could well inspire other regulators, including that of the EU
With generative AI, we are witnessing a standoff between the AI giants and the content creators, media in the lead, whose articles are used to train large language models. The latter believe that AI plunders the fruit of their work, not only by exploiting their texts without paying them, but also by depriving them of income. Indeed, users who access information via ChatGPT, Claude or Gemini simply read the summaries provided by the AI and do not go to media websites, reducing their advertising revenue.
The AI giants, for their part, believe that data publicly accessible on the web must be able to be read and used by everyone, including by algorithms. But also that this data is transformed and not reproduced as is by their algorithms, which extract the logic of the reasoning understood in the texts rather than their content word for word, which does not fall under the laws protecting copyright. And finally that they are content to index the web, as Google has been doing for decades, AI being only a new way of probing the web and accessing content.
Google Zero
However, their position has just suffered a severe snub from the Competition and Markets Authority (CMA), the British competition authority. At the beginning of June, it rendered a judgment against Google, stipulating that it should allow press publishers to unsubscribe from its AI Overviews without disappearing from search results.
Available in the United States, the United Kingdom and in 200 different countries, but not in France, the AI Overviews function constitutes a generative AI insert which appears when you type a search in Google, above the results, and gives a summary of the answer you are looking for. In short, Google Gemini gives the Internet user a succinct answer, before showing them the Google links corresponding to their search.
For more than two decades, search engines, led by Google, have operated under a tacit agreement with the media. They created content, Google helped readers find it, and in return, news publishers received traffic, subscriptions and advertising revenue. The AI Overviews feature makes this balance obsolete.
This functionality also illustrates the dilemma in which press publishers find themselves facing the giant Google. On the one hand, the AI Overviews feature threatens their revenue, since it makes people less likely to click through to their websites to complete their search. On the other hand, asking to be excluded from AI Overviews training algorithms would mean depriving yourself of an increasingly popular way of accessing content, the GEOand expose yourself to retaliation from Google which risks deindexing the website in question in its search results. In short, a choice between slow agony and rapid death.
This is what Nilay Patel, editor-in-chief of the American media The Verge, calls “Google Zero”. Google sucks content from the internet without paying for it, then recomposes it on its own platform, so users never leave.
Why this is an important judgment
“The CMA’s decision is a major step forward. It sets out a simple but fundamental principle: publishers deserve real control over the use of their work to power AI systems. And no company, regardless of its size or power, can claim to control information on a planetary scale,” said Martha Dark, co-executive director of Foxglove, a non-profit organization working to make technology work for everyone. “This decision also validates complaints filed by my organization, Foxglove, and our partners, who have spent the last year arguing that Google abused its dominant position and coerced publishers into an unfair deal.”
“Any new controls must avoid harming search by leading to a fragmented or confusing experience for users. As AI becomes a central part of how people access information, these new controls will also need to be simple and broadly applicable for website owners. We are confident in our ability to find a path that provides even more choice for site owners and publishers, while ensuring users have the most useful and innovative search experience possible. its part declared Google in a press release.
A cryptic formula which suggests that the standoff is not entirely resolved. “The work in the United Kingdom is not finished. We fear that the implementation of the new British rules will not be rapid enough to protect our independent and local press. We will also closely monitor Google’s compliance with these new rules, and the coercive measures that may prove necessary to force them to respect the law”, judges Martha Dark.
A decision of global significance
And this judgment extends well beyond the borders of the United Kingdom. First, the British competition authority exercises strong influence in the rest of the world, in particular on American and European regulators. In 2023, its blocking of the Microsoft/Blizzard Activision merger had, for example, led the FTC and the European Commission to harden their respective positions and obtain more concessions from Microsoft to give the green light to the acquisition. A year earlier, it was his action that led to the forced sale of Giphy by Facebook. Last year, its decision to grant Google strategic market status influenced the application of the Digital Markets Act (DMA) to the online search giant in the EU.
Then, the standoff between Google and press publishers does not only concern the United Kingdom, Google adopting the same strategy everywhere in the world. The online search giant is the subject of similar investigations in Europe and Brazil. “Over the past twelve months, the European Commission’s Directorate-General for Competition and Brazil’s Administrative Council for Economic Defense have both opened investigations into Google’s use of journalistic content in its AI products. The CMA’s decision should give them the confidence to quickly reach the same conclusion,” says Martha Dark.
Especially since Google has just suffered a second snub regarding its AI Overviews functionality, this time from the German regulator. It has in fact ruled that Google could be criminally liable for the false information communicated by this functionality, considering that, unlike traditional search engines which are content to present lists of links to third-party content, Google’s tool formulates “independent, new and substantial declarations” based on its own erroneous interpretation of links found on the internet. A decision which could again be a landmark, by creating a precedent which would mean that the AI giants could no longer be satisfied with a simple warning indicating that the AI can make mistakes to be protected against any legal proceedings.
