Users in Iran used ChatGPT to publish articles in US media, Washington Post reports
ChatGPT users in Iran produced AI-generated articles that appeared in US media outlets, according to The Washington Post. The reported case involves an opinion piece published in the Port St. Joe Star, a small weekly newspaper in Florida. The article discussed the costs of the war with Iran and ran under the byline Ervin Hoskins.
The same name also appeared on articles critical of the conflict in Daily Kos and Middle East Monitor. According to the report, those two publications have more than 3 million Facebook followers combined. That figure reflects the size of their audiences on the social network, but it does not reveal how many people read the articles in question.
The episode raises an important question in the debate over AI and news: how can publishers verify the origin of content when text-generation tools make it easier to write for audiences in other countries? The answer involves reporting, publishing standards, and clarity about who takes responsibility as the author.
What the Florida newspaper published
The article in the Port St. Joe Star made an argument about the economic effects of the conflict. In short, it maintained that the costs to Americans would continue even after the fighting ended. The excerpt provided by The Washington Post does not detail the data used to support that assessment.
It is important to distinguish the content format: this was an opinion piece, not a straight news report. In that format, an author can advocate a position and interpret events. That does not remove the need to check the facts, figures, and sources used to support those arguments.
The Ervin Hoskins byline connects the Florida publication to the articles mentioned in Daily Kos and Middle East Monitor. However, the available material does not establish the identity behind the name or explain how each outlet received the articles.
The same byline appearing in different publications does not, on its own, establish who wrote the articles or how they were distributed. Nor does it establish that editors knew about the reported location of the users or the use of AI to produce the content.
What the report says about the use of ChatGPT
The Washington Post attributes the AI-generated articles to ChatGPT users in Iran. That is the central finding of the report. However, the available excerpt does not include the records, verification methods, or other evidence the newspaper used to reach that conclusion.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
That distinction needs to be clear. The absence of those details from the excerpt does not invalidate the reporting, but it limits what can be explained based on that excerpt. The material does not provide a complete account of how the articles were created and submitted.
Where artificial intelligence can play a role
Tools like ChatGPT can assist with several parts of the writing process. They can organize ideas, prepare initial drafts, translate content, and adjust the tone of a message. They can also revise sentences for clarity. These are general capabilities of the technology, not confirmed steps in this particular case.
For that reason, the phrase AI-generated articles does not, by itself, indicate how much human work went into a piece before publication. A text may undergo extensive editing or be submitted with almost no changes. The report excerpt does not establish which of those situations occurred.
The distinction matters when assessing the role of the tool. Still, in any scenario, responsibility for verifying the content and deciding whether to publish it remains with the people involved in that process.
What cannot be attributed to the users
The reported location of the users does not establish a connection to a government or organization. The supplied material also does not detail funding, shared instructions, or a coordination structure among those responsible for the articles.
Presenting the episode as a confirmed state-run operation would therefore go beyond the available information. The same applies to claiming that all content published under that name was produced using the same method or submitted by the same person.
Social media reach is not the same as article readership
The reference to more than 3 million followers across Daily Kos and Middle East Monitor helps illustrate the size of their Facebook presence. It does not, however, represent a count of readers of the Ervin Hoskins articles.
Followers, views, and article reads are different metrics. Someone may follow a page without seeing a particular post. They may also see an article preview on social media without opening the article. In addition, some people may follow both outlets.
The available excerpt provides no basis for claiming that the articles reached 3 million people. Measuring their reach would require specific data, such as page views, post impressions, and shares.
The case involves a small local weekly and publications with larger social media followings. That contrast shows the variety of outlets mentioned, but it does not establish a deliberate strategy to reach different audiences.
AI-generated text is not automatically misinformation
The use of artificial intelligence deserves scrutiny, but it is not enough to classify an article as false. Content created with help from ChatGPT can contain accurate information. Likewise, a text written without the tool can contain errors, data taken out of context, or weak sourcing.
An assessment needs to separate three issues: the accuracy of the claims, the identity of the person named in the byline, and transparency about how the content was produced. These aspects are related, but they are not interchangeable. A text can present accurate data while still raising meaningful questions about its authorship.
It is also necessary to distinguish an opinion from a factual claim. Arguing that a conflict will have lasting economic effects is a position that can be debated. Citing a figure for government spending, on the other hand, requires a source, a time frame, and an explanation of what was included in the calculation.
The excerpt from The Washington Post does not provide a detailed fact-check of the claims in the articles. It therefore does not support a conclusion that their arguments were false simply because they were linked to the use of ChatGPT.
Persuasive writing is no substitute for evidence
Language models generate responses based on patterns learned during training. This allows them to produce natural-sounding sentences and well-organized text. But fluency does not guarantee accuracy: these tools can present incorrect information in a confident tone.
In practice, an article may look ready for publication and still require careful review. Names, dates, figures, and references need to be checked against appropriate sources. Polished writing does not eliminate that step.
Why authorship matters in outside contributions
Articles submitted by outside contributors broaden the range of voices in a publication. At the same time, they require standards that help readers understand who is speaking and which affiliations may be relevant to the opinion being expressed.
A byline is more than a name at the top of a page. It helps establish responsibility for the claims and gives readers a way to learn about the background of the author. When that information is vague or cannot be verified, editorial assessment becomes more difficult.
Without assuming that the publications mentioned fell short, a few general safeguards help explain what this work involves:
- Verify the contributor: confirm basic information and maintain a reliable way to contact the person who submitted the material.
- Understand relevant affiliations: identify professional relationships or interests connected to the topic.
- Check factual claims: review figures, dates, names, and references, including in opinion pieces.
- Set rules for AI: clarify which uses are acceptable and when they must be disclosed.
- Maintain editorial accountability: do not treat fluent writing as proof of thorough reporting.
These measures are useful even when artificial intelligence is not involved. The technology makes the discussion more urgent because it can reduce the effort needed to write and adapt content, including in another language.
What this case adds to the debate over AI and news
The episode reported by The Washington Post shifts attention from writing quality to the origin of the content. The issue is not just whether a text sounds human, but who produced it, how it reached the outlet, and what verification supported its publication.
The available material describes the reported use of ChatGPT by users in Iran, an opinion piece published in Florida, and a byline that also appears in other outlets. There are not enough details to measure the readership of the articles, establish institutional ties, or assess the entire production process.
The central editorial issue is trust in authorship and information. AI tools can assist with writing, but they do not replace verifiable sources, clear standards, or accountability for what reaches readers.
