Hegseth attacks CNN after exposing a disinformation campaign linked to Iran

Listen to the article. The audio text is automated, generated by an automated system

US Secretary of Defense Pete Hegseth attacked CNN, after it published a report on entities operating from Iran using artificial intelligence tools and fake journalistic identities to publish articles criticizing the US war on Tehran in media organizations.

Hegseth wrote, in a post on the “X” platform, addressing the network: “You fall into the trap over and over again, because you hate the president more than you love the country and our soldiers.”

The US Minister's comment came in response to a report published by CNN on Thursday, based on results announced by OpenAI regarding a media influence operation whose models were used to prepare political content and promote it through the media and social networking sites.

According to the CNN report, OpenAI revealed that about 12 media organizations around the world published nearly 100 articles attributed to 7 fake journalists, as part of an activity that used artificial intelligence to influence the public debate regarding the war with Iran.

The company explained that it banned accounts on “ChatGPT” that were used by those in charge of the activity to reformulate articles and proposals for publication in the Persian language, as well as to generate comments intended for social media platforms.

Some articles dealt with American policy towards Iran, the role of Jared Kushner in foreign policy, and the credibility of Western positions during the war.

The network indicated that the institutions that published the articles do not appear to include major US media outlets with widespread circulation, but some of them have a large following on the Internet.

Among these institutions is the Middle East Monitor website, which is based in London and has about two million followers on Facebook.

The site's director, Daoud Abdullah, told CNN that he will examine the articles referred to by OpenAI and will take appropriate measures.

The American “LA Progressive” platform also published an article that discussed Kushner and American foreign policy, while “CNN” said that it had requested a comment from the platform.

The process was not limited to publishing articles, as OpenAI reported that account users asked ChatGPT to create comments that appeared to be coming from real readers interacting with the published materials.

The proposed comments addressed missile strikes, ceasefires, and diplomatic moves, with the content directed to portray the United States as the aggressor, Israel as an ally that cannot be trusted, and Iran as a resilient victim.

CNN quoted Jessica Brandt, former official in charge of the Center for Countering Malicious Foreign Influence at the US National Intelligence Agency, as saying that the arrival of this content to actual media institutions represents an important development in the methods of influence operations.

She explained that artificial intelligence allows the parties carrying out these campaigns to hide their identity in a more complex way and use real media channels to expand the spread of their messages.

Although the activity was linked to Iran, OpenAI stressed that it was unable to conclusively determine the party responsible for it, or prove its connection to the Iranian government.

The company said that the nature of the activity appears consistent with an influence operation carried out by a commercial entity on behalf of another party, but it was unable to determine the identity of the entity concerned.

This means that the published results do not prove that the Iranian Revolutionary Guard or any Iranian government institution directly managed the operation.

This issue comes amid the increasing use of artificial intelligence tools in war-related media campaigns, whether producing articles under pseudonyms or preparing video clips and comments aimed at directing public opinion.

CNN had previously spoken about the use of artificial intelligence to produce satirical clips related to war, as well as the dangers of misleading intelligence information that is created or processed using these technologies.

The case highlights a growing challenge for media organizations in verifying the identities of writers and the sources of materials submitted for publication, especially when artificial intelligence tools are used to give a professional appearance to content aimed at political influence.

While Hegseth chose to attack CNN based on the report, the identity of the party that ran the campaign and the extent of its connection to the Iranian authorities remain issues that have not been resolved by OpenAI’s announced results.