By Alessandro CONTE
Thirty-eight pro-Russian articles, produced with ChatGPT and published on South African websites under fictitious names. It is one of the episodes reconstructed in the 2026 report, dedicated to the weakness of information. A limited but significant case: to conduct an influence campaign, you no longer need large clandestine editorial offices. All you need is access to an artificial intelligence system, a few intermediaries and a network of platforms willing to publish content without verifying its origin.
The report places these phenomena in the context of “foreign information manipulation and interference”, known by the acronym FIMI, as well as influence operations carried out by foreign actors. Disinformation thus becomes a component of hybrid warfare: it accompanies economic pressures, cyberattacks and hostile diplomatic initiatives, with the aim of orienting the debate, delegitimizing institutions and deepening internal divisions.
Forbidden Stories’ investigation into South Africa has documented payments to intermediaries tasked with promoting narratives favorable to Moscow, through influencers and local media outlets. Artificial intelligence has provided the raw material, speeding up the production of seemingly journalistic texts. This is where the real difference lies: a single person can staff multiple sites, simulate different identities, and tailor the same message to different audiences, at minimal cost.
Election campaigns are the most sensitive terrain. During the 2026 Japanese snap election, according to a Nikkei investigation, 55 percent of political video views were generated by anonymous creators. The lack of a verifiable identity makes it impossible for voters to reconstruct the interests, funding, and international connections behind them. A video can reach millions of people before authorities, platforms, or journalists can determine who produced it.
In Hungary, the manipulation has taken an even more aggressive form. Deepfakes and defamatory videos against the opposition leader have accumulated around 650 million views on YouTube and Facebook. Their circulation has also been reinforced by government-affiliated media outlets. Thus, digital forgery has transcended the boundaries of social networks, entering the traditional media circuit and gaining a semblance of legitimacy.
The same dynamic plays out in armed conflicts. During the border clashes between Thailand and Cambodia, various platforms were flooded with out-of-context images, unsourced reconstructions, and nationalist messages. Creators and self-styled “citizen journalists” have filled the space left empty by official sources, turning every military episode into material to fuel fear and hostility. In a crisis, speed often rewards the most emotional content, not the most accurate.
Deepfakes and “AI slop,” the flood of low-quality, mass-produced synthetic materials, further complicate the work of editorial staff. In Western Europe, concern about the political use of artificial intelligence has increased by nine points in Belgium and eight in the Netherlands. The problem is not only related to the quality of the fakes. Above all, their volume matters: verifying one photo takes time, while generating a hundred of them takes only a few minutes.
Some media outlets are building special structures. Danish broadcaster DR has launched “DR Verify”, Ireland’s RTÉ has created the “Clarity” service, while Thai PBS has been using the “Verify” platform since 2025, used during tensions on the Cambodian border. In Taiwan, public service TBS is experimenting with a chatbot capable of linking responses to authoritative journalistic sources. The BBC and Swedish media, meanwhile, insist on the “human in the loop” principle: no AI-assisted content should be published without editorial control.
The response also includes political agreements and regulatory tools. In Norway, the main parties have committed not to use deepfakes during elections. In Spain, El País is collaborating with Maldita.es, while the government has announced the creation of HODIO, an artificial intelligence-based system to monitor the spread of hate and polarization online. Singapore has chosen the legislative route with the Online Safety Bill, expanding protection against harm caused by manipulated content.
However, a structural weakness remains. Global trust in news has fallen to 37 percent, while so-called “Google Zero” is reducing direct traffic to news sites: synthetic answers built into search engines keep users inside them, without taking them to the original sources. As a result, the public is faced with less verified content and more narratives distributed by anonymous creators, automated systems and coordinated networks.
A less grim signal comes from chatbot users: 44 percent click on suggested links to check the credibility of answers. It’s a useful skepticism, but not enough. The next frontier will be the ability to distinguish sources, authors, and synthetic content.
If this traceability remains optional, artificial intelligence will continue to offer influence operations exactly what they need most: speed, anonymity, and instant access to already divided public opinion. (European Affairs)