What if, in addition to producing much faster and for much less, you could still be at the top of Google searches and receive a large volume of visits generating tens of thousands of dollars in advertising revenue? The scenario above challenges common sense, but it is a growing reality.

Thanks to ChatGPT and other generative AI tools, an industry is rapidly growing that profits from copying and rewriting content from reputable news sites. The logic is the same: identify the most-read stories on the internet and use AI to rewrite these texts and optimize the content to appear at the top of Google platforms. The more publications, the better.

The problem is that those who invest in creating original and high-quality content earn nothing and still lose readers and advertising to this competition that copies them. For comparison, the minimum wage for a journalist in the state of São Paulo, one of the highest in the country, is R$4,275.75 for a five-hour shift—approximately $804.13—(with charges, the cost to the company practically doubles).

Assuming that the journalist costs only R$4,275.75, with AI-generated texts already costing less than R$2.15, a journalist would have to produce nearly 2,000 texts per month to “pay yourself” for competing with AI (considering the charges, the number would jump to almost 4,000).

Meeting Google’s demand

Newzstock is one of the companies offering an AI tool to identify the most-read news, copy, rewrite, and publish it on websites. According to Pedro Laganá, one of the company’s owners, the idea for Newzstock came from publishers’ need to produce more and more content to appear on Google.

“A writer will spend two to three hours to find something new [to write about],” said Mr. Laganá. “After they find it, they’ll spend more time adapting this text to their website. That’s where the idea for my platform came from. First, it’s a news clipping platform. You find it by category, keyword, city, or language. Then, a secondary feature is the ability to rewrite the text.”

Mr. Laganá’s platform scans over 300,000 news websites worldwide and highlights what is “doing well” on Google. Then, a writer inserts the text link into a panel that rewrites the text (or translates it if necessary), and the journalist can add or remove information or simply publish what the platform’s AI produces. In a few seconds, the new version of the text is available and can be published on the site with just one click. Newzstock uses OpenAI’s ChatGPT to alter and translate the texts.

I asked Mr. Laganá if the 300,000 websites scanned by the platform authorize the practice and if they are somehow paid. “The main responsible for the information is the API I use.” API is a set of rules that allows communication between different computer programs. In other words, from Mr. Laganá’s perspective, the company he pays to receive this data is responsible for making agreements with the publishers. “We have this sensitivity because some big publishers use our platform.”

He also says that it is up to editors to use the content correctly. “If you simply copy and paste, you will be acting against Google’s policies and against yourself,” Mr. Laganá said.

“We saw a problem with a large publisher a few weeks ago where they hired a consultant who was supposedly going to help create content, but he just entered ChatGPT and created a fictitious journalist and started publishing fictitious articles. My point is to give more agility to the journalist and help them find content to be more productive.”

Mr. Laganá also emphasizes that Newzstock is not a publisher, just a tool, and the final responsibility lies with the publisher. According to him, no publisher has asked to be removed from the platform so far.

Google’s silence and frustrated journalists

Mr. Laganá, who is 28 years old and holds a marketing degree, said he first encountered the editorial market six years ago when he started working with website advertising. Because of his experience, he established a policy of requiring source citations for investigative reporting and exclusive news. “We curate to ensure that publishers are credited.”

A common point among Newzstock’s clients is that, even using AI in their texts, this information is not revealed to readers.

Mr. Laganá’s frank manner of talking about his company is surprising, but it is understandable that he sees no problems. The reality is that, in this market, Newzstock is among the less aggressive AI tools. There are dozens of sites and content agencies that just copy, rewrite, and republish automatically without even using humans in the process.

Many also resort to false and sensational titles to get more clicks and become MFA sites by filling their pages with ads, deceiving advertisers. There are also content consultancies that publish identical texts on hundreds of sites, many of them large outlets, or create everything with ChatGPT.

Publishers heard by the column, who spoke on the condition of anonymity, say that AI platforms indeed speed up the work and that there is considerable audience growth from Google, even if Google’s recommended practices are not followed.

“Today, everyone copies everyone. It’s a disgrace, we have to do this or disappear on Google and die. I wanted to produce more original content; do you think any journalist studies for four years to pressing copy and paste buttons? This is even causing a problem in retaining talent.”

Journalists heard by the column say that since Google dominates more than 90% of searches, controls Android phones—which are predominant in the market—and has more than 70% of internet users using Chrome, it is virtually impossible to escape its monopoly.

“We strive to produce original, high-quality content,” said Daniel Castro, publisher of the site Notícias da TV. “The merit/audience should be for those who break the news. It’s the pillar of journalism; it’s what drives reporters and some editors. This should be respected. It would be a basic and easy rule to follow. Whoever breaks the news first should be highlighted in Google results. Unfortunately, that logic does not exist. The audience is very scattered, and content made by robots or writers without investment in investigation and journalism is dominating.”

The column contacted Google’s communications department on May 21 and April 30 and sent questions related to possible punishments for sites that use AI tools and why copied content ranks higher than original reports, but received no response. If they respond after the publication of this text, the answer will be added at the end of the text.

Journalism is just the tip of the iceberg

“Law always follows technology, trying to regulate and establish a balance in these relationships,” said Camila Gurgel Guglielmo, a master’s degree holder in Intellectual Property Rights from Université de Montpellier. “On one side, there are those who have spent time, knowledge, and investment to create an original work. On the other, technological advancement, which cannot stop. AI brings the whole issue of attribution of rights into discussion.”

Large language models are created from information and data that have intellectual property, so there is a discussion about whether the results created by AIs are derivative works or not. But Ms. Guglielmo emphasizes that we have strong copyright laws in Brazil, and indeed, there is the right to use small excerpts, but if it becomes a copy or a derivative work, there is, in theory, a violation of copyright.

In Ms. Guglielmo’s view, sites and editors that use AI and data from other sites also assume risks, especially if jurisprudence is formed. She cites the case of Getty Images, which is suing Stability AI for unauthorized use of images, as a potential precedent to be set soon.

The problem is that, in practice, the law is not working. Several content producers told the column they are facing growing problems with copies and unauthorized use of intellectual property. After all, content copying continues to grow dramatically, including for AI training, and the legislation is becoming outdated in the face of new challenges created by technology. To make matters worse, the influence of Big Techs in Brasília has prevented any effective change in the sector.

Therefore, how the issue of rampant use of journalistic content is addressed by lawmakers and authorities can also determine the understanding of AI use in various sectors. And it is no secret that Google and Meta, two protagonists in AI, refuse to pay for journalism.

“AI tools that create websites with third-party content without authorization commit a clear and obvious copyright violation,” said Marcelo Rech, executive president of the National Association of Newspapers (ANJ). “Newsguard has already identified more than 800 new sites, including in Portuguese, with little or no human intervention, that copy content, shuffle it, and distribute it for free to profit from clicks.”

ANJ has issued a series of guidelines to its members to try and block the illegal use of journalistic content, as well as recommending explicit disallowance of capturing such content without prior agreement or agreed remuneration.

“An AI tool, whether the pseudo-journalistic sites or LLMs that capture this content without authorization, is equivalent to entering people’s homes and taking all the furniture, paintings, rugs, and belongings. Claiming that the door was not properly locked is morally unacceptable. That is why journalistic organizations like The New York Times and Getty Images are already taking action against LLMs”.

Regina Bucco, executive director of the National Association of Magazine Editors (ANER), said that the association “encourages the learning and use of new digital tools as long as they facilitate research, investigation, fact-checking, and organization or analysis of large volumes of data and any activity that enriches the quality of journalism,” but adds that “as an association that defends serious and professional journalism, we do not agree with the use of AI or generative AI to plagiarize, copy content, or for mass distribution without reference to the sources of production.”