Two stories, one question
Two events from this week deserve to be read together.
The first: Donald Trump posted on Truth Social accusing major American media outlets of spreading false information about the war in Iran, suggesting that those responsible could be “brought up on charges for TREASON for the dissemination of false information.” He also praised his FCC chairman, Brendan Carr, who was threatening to pull the licenses of broadcasters producing what he calls “hoaxes and news distortions.”
The second: the Financial Times, the Guardian, the BBC, the Telegraph and Sky News announced their intention to form a coalition — which they themselves call an “information NATO” — to force AI companies to pay for the journalistic content used to train their models. Norwegian media are going further: they are funding their own language model (LLM) from national archives.
These two stories seem unrelated. Yet they speak to exactly the same thing: who controls what is true, and who profits from it.
What is established
On March 16, 2026, Trump published a 401-word Truth Social post accusing the media of coordinating with Iran to spread false information — in particular, an AI-generated video showing the USS Abraham Lincoln aircraft carrier on fire. He suggested “TREASON” charges.
CNN fact-checked the claim: no major American media outlet had distributed the fake video. The only references found pointed to three foreign outlets — one Israeli, one Turkish, one Saudi. When the White House was asked for specific examples, it provided none from American media.
The maximum penalty for treason in the United States is the death penalty.
FCC Chairman Brendan Carr threatened broadcasters with losing their licenses. Media law experts called this threat “hollow” — the FCC has not revoked a license in decades, and any legal action would trigger a lengthy process protected by the First Amendment.
On the other side, generative AI systems have indeed been built using substantial volumes of journalistic articles freely accessible online, without compensating the content producers. This is documented and not contested by the AI companies themselves.
Patrick Bourbeau, Vice President of Legal Affairs at La Presse, confirmed that web crawlers are actively circumventing the blocks put in place by media outlets to prevent this type of extraction.
What is debated
Are Trump’s threats against the media serious or rhetorical? Legal experts broadly agree that the current threats are unlikely to result in actual prosecutions — constitutional protection of press freedom is robust. But Jameel Jaffer, executive director of the Knight First Amendment Institute at Columbia, sees in them “an intensification of a long-standing effort to bring news organizations into closer alignment” with the president’s political agenda. Pressure does not need to be legally viable to produce real chilling effects on editorial freedom.
Can the European coalition change the rules? Bourbeau believes that if the coalition succeeds in “turning off the tap,” it could force AI companies to change a model in which they currently “pay nothing for content.” Other observers are more skeptical: major platforms have the legal and political resources to hold out for a long time, and some AI models now use synthetic data that bypasses dependence on journalistic content.
Is the Norwegian model transferable? Norway is funding its own LLM from national press archives with state support. Professor Pierre Trudel of the Université de Montréal believes that “launching initiatives like the one by European media is a good way for these companies to regain some control over the challenges posed by the generalization of AI.” But a national LLM requires considerable resources, sustained political will, and raises its own questions about state control of information.
What remains hypothetical
If a government can label as “fake news” anything that disturbs its official narrative — and threaten prosecutions against those who publish otherwise — then the question of who decides what is true is no longer academic. It becomes a matter of survival for independent journalism.
It is too early to say whether the European coalition will succeed in imposing a viable compensation model. It is equally impossible to predict whether Trump will go beyond rhetoric. Recent history suggests that pressure, even without direct legal sanction, produces lasting effects on editorial freedom.
What these two stories reveal — our position
There is a deep irony in major media outlets demanding an “information NATO” to defend their content against AI — at the very moment when those same outlets face a credibility crisis that AI did not create, but that it amplifies.
The real question is not “how do we stop AI from using our content?” It is “why do millions of readers no longer know whether what they read is true?”
Trump can threaten the media with treason because enough people have lost the tools to distinguish a verifiable claim from a political opinion presented as fact. This confusion is not the result of a conspiracy — it is the result of a decade of journalism that gradually erased the boundary between established information, debated interpretation, and speculative hypothesis.
What we propose is not a defense of the media, nor an attack on them. It is a method.
Explicitly distinguishing what is established (verifiable, sourced, uncontested), what is debated (legitimate interpretations in conflict), and what is hypothetical (reasoned speculation but not yet verifiable) is not a revolution. It is a return to the foundations of rigorous journalism — applied transparently, visible to the reader.
This is what HUMANITY.NET practices in its own articles. This is what Hum_ID allows every user to demand from the AI they use.
And it is, we believe, the only lasting answer to a world where power has learned to weaponize doubt.
Real or not real? The facts cited in this article are sourced and verifiable. Our conclusions are clearly identified as our own.