The world's first AI newspaper and the question that makes the press ponder

(Dan Tri) - More than a year since the world's first printed newspaper powered by AI was born, the debate has not cooled down in the context of AI application causing much controversy.

The editor-in-chief's strange conversation

- Hello

- Hello, how are you?

That was the opening paragraph in the summary of the Foglio AI project, when Claudio Cerasa, Editor-in-Chief of the Italian daily newspaper Il Foglio, decided to let artificial intelligence interview him about the time the two sides worked together.

Claudio Cerasa, Editor-in-Chief of Italian daily newspaper Il Foglio, participated in an interview with artificial intelligence to summarize the Foglio AI project (Illustration: Loyal Source).

During a long conversation with tens of thousands of words, the AI ​​continuously asked questions about journalism, the future of journalism, and itself.

Sometimes the chatbot is humorous, sometimes sarcastic. However, there are also moments when AI resembles an intern trying to prove its worth to the manager.

The conversation made many people feel like they were seeing an image of a future newsroom, where editors and AI sit around a meeting table to exchange work ideas.

Everything actually started in March 2025, when Il Foglio released a four-page supplement announcing that all content, from titles, editorials, comments, reader letters to summaries, was created entirely by AI.

According to Cerasa, this is the first paper newspaper globally made entirely by artificial intelligence.

For many years, newsrooms around the world have quietly tested AI for many different tasks such as translating articles, suggesting titles, searching for documents or summarizing texts. But Il Foglio decided to go further by bringing AI to a position that is considered central to journalism, which is article writing.

The editorial board is responsible for providing topics, approaches, perspectives and specific requests. AI will take care of writing the article, setting the title, choosing the featured quote, and creating the summary. Then, editors will review the content before publishing.

It is worth noting that Cerasa never saw this project as a way to replace journalists. He sees it as an experiment to better understand how technology is changing journalism.

"AI is like the wind. You can't hide it. You have to learn how to control it," Cerasa told The Atlantic.

The project ended after a month of testing. Instead of becoming a short-lived experiment, Foglio AI opened up a debate that continues to this day.

Because what Il Foglio discovered is not just the capabilities of AI. More importantly, the experiment forced many people to reconsider the nature of journalism in the face of the wave of technology.

The test results went against expectations

Initially, Cerasa was really surprised by AI's capabilities. At the time of launch, the first issue of AI also helped Il Foglio's newspaper sales increase by about 60% on the day of release.

AI can read documents hundreds of pages long in just a few minutes, summarize speeches by politicians, analyze books, simulate the style of various authors and even create fictional debates between characters who have never met each other.

And what surprised him the most was the AI's ability to be humorous and sarcastic.

Before starting the project, he did not believe that a chatbot could write such ironic paragraphs.

AI's ability to be humorous and sarcastic is one of the things that surprised Editor-in-Chief Il Foglio the most (Illustration: Getty).

However, during the testing process, the AI ​​surprised the editorial board many times with unexpected expressions.

However, the longer it is used, the more clearly the limitations of this technology become apparent.

According to records, there have been many times when AI mistakenly thought Donald Trump was still in the US presidential election campaign even though he had returned to the White House. There are times when AI cites events that have never happened or creates completely bogus facts.

Sometimes, this "virtual journalist" even comes up with attractive headlines, but when reading the content, it is not related to the headline itself.

These limitations are also what many journalism experts point out.

In a post on the website of the Poynter Institute, an American non-profit organization specializing in journalism training, Alex Mahadevan, director of MediaWise, commented that Foglio AI is a bold experiment but at the same time shows that AI still has many weaknesses when participating in news production.

Mahadevan cited the findings of Alexios Mantzarlis, director of the Security, Trust and Safety Initiative at Cornell Tech, in which some articles were found to have informational errors, lack evidence and use stereotyped endings.

However, what makes Claudio Cerasa really ponder the most is not the errors mentioned above but the difficult limits of AI.

Chatbots can process huge amounts of information quickly, can write quickly, and sometimes quite well. But it doesn't really understand what a story is worth pursuing.

AI cannot detect an exclusive topic on its own, cannot build a network of news sources, and moreover cannot recognize an unusual detail in a city council meeting and then follow that trace to open an investigation.

It also cannot sense the atmosphere of an event, the facial expressions of a character or the hidden signals that appear during an interview. Most importantly, according to Cerasa, AI does not know how to argue.

He made a brief comment on the difference between journalists and artificial intelligence: “AI doesn't sweat and doesn't argue.”

A newsroom often operates through debates, disagreements, criticism and mutual correction.

AI's fatal weakness lies in its ability to debate and criticize, elements considered the core of journalism (Illustration: Scrabbl).

A reporter may object to an editor's idea. An editor can overrule a reporter's approach. Just one unexpected phone call from the source can cause the entire structure of the article to change.

It is from those collisions that important journalistic stories are formed. And of course, AI doesn't do that, it just responds based on the questions and requests given.

From these very perceptions led Cerasa to a somewhat paradoxical conclusion. After a month of testing the AI ​​newspaper, he believes in the future of journalism more than before.

Answers from the first AI newspaper

The story of AI newspaper testing continues to be debated at a time when the global press is recently facing unprecedented pressure from AI.

In the US, the Plain Dealer newspaper in Cleveland recently caused controversy when Editor-in-Chief Chris Quinn revealed that a new graduate withdrew from the internship program when he learned the job was no longer focused on writing articles. Instead, interns mainly go to the field, collect information and send notes to the AI ​​to draft.

He said that letting AI take over part of the writing work gives reporters more time to contact sources, attend meetings and pursue more important stories.

“Advance Local Express Desk” is a new pseudonym appearing on the Plain Dealer newspaper to mark articles edited by AI (Screenshot).

This viewpoint has faced a wave of fierce opposition in the press. Former Financial Times Editor-in-Chief Lionel Barber said this way of thinking was a rash decision.

Meanwhile, reporter Sam Allard commented that the intern candidate simply wanted to become a journalist, not someone who provides raw materials for AI production. Editor Philip Lewis even commented that an editor-in-chief who encouraged the elimination of reporters' writing jobs should resign.

Meanwhile, at a meeting with employees, Mathias Döpfner, CEO of Axel Springer, is said to have quite similar views to Quinn when declaring: "Either accept AI, or die."

In an op-ed for The Guardian, journalist Margaret Sullivan pointed to the controversy at the Plain Dealer as an example of how journalism is still debating the role of AI in the news production process.

While not taking sides, Sullivan emphasized that embracing AI does not mean turning all journalism over to machines.

As AI becomes more and more present in newsrooms, the issue is no longer whether to use it or not, but rather determining the boundaries between what AI can do and the tasks that need to be done by humans (Illustration: AI).

According to her, the most important principle is to maintain "human control", meaning that all content created by AI still needs to be checked, verified and held accountable by reporters and editors before publishing.

She pointed out that AI has proven its value in many large-scale investigation projects, helping to process huge volumes of data that humans can hardly do in a short time.

But in the opposite direction, a series of incidents related to fabricated content, false information or mass-produced products using AI also show the risks if this technology is abused. Perhaps that is also the lesson that Foglio AI leaves behind.

After the end of the AI ​​print newspaper test in 2025, Il Foglio continues to maintain the "Il Foglio AI" section on the website, where articles and comments are published entirely using artificial intelligence (Screenshot).

After experiments and observations from articles written by AI, what is more interesting is what are the limits of technology and what are the tasks that journalism still needs humans to undertake.

Because if AI becomes increasingly proficient in creating words, the value of journalists will lie in the ability to discover stories, build sources, understand social context and have critical thinking. Those are capabilities that, at least for now, machines cannot replace.