“A model that interacts with users as if holding a conversation. ChatGPT can answer users’ clarifying questions, admit mistakes, question premises it considers incorrect, and reject inappropriate requests.”
—OpenAI, 2022

On November 30, 2022, ChatGPT was officially launched on OpenAI’s website. The page stated that the new tool was capable of following instructions and providing detailed responses.

Nearly three years after this milestone, the impact of this technology on journalism remains an evolving story. While its integration has been gradual across different markets and formats, the media industry now has multiple examples of AI being applied in various processes of the management, creation and distribution of content.

In 2024, The Washington Post launched an experimental tool that allows users to ask questions and receive answers based on its published journalistic content. In Latin America, the Argentine media outlet Clarín introduced UalterAI, a resource capable of creating summaries and timelines and of providing information in a question-and-answer format.

Michael Krisch, interim director of the Brown Institute at Columbia Journalism School, stated, “We’re seeing much greater incorporation of AI across the various workflows of both journalists and editors, as well as many business leaders. So yes, I would say it touches all facets of journalism at this point.”

The Associated Press (AP), one of the world’s most influential news agencies, has begun exploring the use of artificial intelligence to optimize news production and improve editorial efficiency. Among the tools they have implemented are an automated system for translating articles from English to Spanish, an AI for analyzing audiovisual content and a generative pilot that drafts headlines and summaries.

The report The Real Value of AI for Publishers Today by the World Association of News Publishers (WAN-IFRA), published in October 2025, revealed that over 75% of news organizations reported improvements in their workflows. Additionally, 64% experienced an increase in content production capacity and 55% managed to reduce the time required to publish.

Krisch stated that there are cases where AI applications are “fantastic” and referred to the Scorching Cells report published by Reuters, in which a tool was used to code the data.

“No one is going to sit down and review 3,000 Freedom of Information Act (FOIA) requests from prisons to manually process temperature records. No newsroom could afford to pay a researcher to do that,” he said.

Economic challenges

Although the WAN-IFRA report confirmed the value of AI in operational efficiency, only 9% of the surveyed media outlets reported direct revenue increases attributable to the use of artificial intelligence. Furthermore, 13% reported a “significant or high” value of these tools for monetization.

Krisch acknowledged that deploying a customized AI system requires certain technical expertise. “That demands a team and that team is expensive. Therefore, smaller newsrooms have less capacity to carry out that type of work.”

The Reuters Institute conducted a survey of more than 300 media executives and revealed that, although many participants planned to close lucrative deals with AI platform licenses, there is limited optimism regarding the return on investment.

One-third of participants (35%) stated that the largest portion of revenue will go to major media companies and around half (48%) expressed that, in the end, there will be very little money left for publishers.

On the other hand, according to a report by the Gabo Foundation, Latin America is filled with “information deserts,” meaning the conditions for practicing journalism professionally and consistently are critical. In Chile, 47.5% of the country is considered an information desert, followed by Argentina with 43% and Mexico with 42.5%.

At the same time, the report The State of Local News 2025 from Northwestern University in the United States revealed that more than 50 million people live in counties without their own news outlets or with only a single news source. In parallel, 136 newspaper closures were recorded in the past year in the United States.

Michael Krisch was asked about the impact of artificial intelligence on local media and said that the effect of various public AI systems, such as OpenAI, Anthropic, or Google’s Gemini, is “affecting search traffic that would typically bring people to a news outlet’s website, and suddenly, without traffic, subscribers are lost. Without subscribers, a news organization is lost,” he explained.

The Tow Center for Digital Journalism report, Zero Journalism: How Platforms and Publishers Are Navigating AI, emphasized that the biggest danger posed by artificial intelligence is generative search and tools that summarize news directly on platforms. These systems prevent users from clicking through to visit the media outlet, which results in less traffic on the websites.

“It will reduce certain tasks that journalists have to do. And if that means they can go out and do more reporting, then yes, it could help expand capacity. But it certainly won’t solve the closures of local news outlets,” concluded the interim director of the Brown Institute.

Despite the previous figures, the WAN-IFRA report highlighted that 62% of respondents anticipate that artificial intelligence will provide significant or high value to monetization strategies in the future.

WAN-IFRA identified several areas where AI is already generating a strong return on investment. A Norwegian international media group, Schibsted, boosted subscription sales on its homepage by 75% using an AI model to deliver personalized content in real time to anonymous users.

What about credibility?

Michael Krisch said that one of the biggest challenges of using AI tools that are “facing the public” is trust. “No AI system is perfect. For an industry whose currency is trust, if you violate that trust even once, you lose credibility. And potentially, you can lose a large audience,” Krisch said.

The interim director of the Brown Institute at Columbia emphasized that it is important for media organizations to ask what their goal is in using an AI model. He also believes that safety boundaries must be established around the system so that when it fails, there are mechanisms to address the problem.

The 2025 National Report on News Consumption and Journalism Evaluation in Chile, by the Pontificia Universidad Católica de Valparaíso (PUCV) and Feedback, revealed that 76.2% of respondents believe that the use of AI in media will lead to an increase in the production and dissemination of false news. Along the same lines, only 15.6% of the audience would be willing to consume news generated entirely by AI.

The Zero Journalism report detailed that interviewees considered the technology too unreliable to risk losing trust by publishing content invented by artificial intelligence.

During the interview, Krisch recalled an American newspaper that published a list of the best books to read in the summer: “It was clearly written by AI because some of the books didn’t exist. It was a big mistake for the newspaper,” he said.

Despite all the challenges, Michael Krisch is not opposed to the use of artificial intelligence and believes it can be helpful for journalists. “But there are also many obstacles. If people don’t know how to use the tools responsibly, they can get into trouble very quickly.”

By Valentina Rojas Legisos