Summary
Imagine you open your phone one morning and see a breaking news video.
A world leader is on screen, speaking in their own voice, announcing something shocking — a ceasefire collapse, a missile launch, a confession of wrongdoing. Millions of people share it within hours.
Politicians react. Stock markets move. Then, two days later, the truth comes out: the video was entirely fake, generated by an AI system in minutes. No camera. No studio. No real person. Just code.
This is not a future scenario. It is happening right now, and it is one of the most serious challenges facing countries that value democracy, truth, and free elections.
We are living through what experts call cognitive warfare — a form of conflict that does not use bombs or soldiers but instead targets the human mind.
The goal is to make people believe things that are not true, feel emotions that push them toward certain decisions, and lose trust in the institutions and leaders that hold society together. And today, powerful AI systems called foundation models are making this kind of warfare easier, cheaper, and more dangerous than ever before.
Foundation models are large AI systems trained on enormous amounts of text, images, audio, and video.
They are the technology behind chatbots, image generators, and voice-cloning apps. They were built for helpful purposes — answering questions, writing emails, generating pictures for creative projects.
But the same tools that help a student write a school essay can be used by a government spy agency to produce one thousand fake news articles overnight, or to create a convincing video of a politician saying something they never said.
In early 2026, during the conflict involving Iran, researchers identified more than one hundred and ten unique AI-generated videos spreading false stories of Iranian military victories.
Fake images showed rockets hitting cities that were never attacked. Fake videos depicted downed aircraft that were never shot down.
These were created by networks connected to the Iranian government and then amplified by Russian and Chinese online systems — each playing their part in spreading confusion across the world.
Russia has been doing this in Europe for years, but the tools have become far more powerful. A British university professor recently discovered a video of himself — made entirely by AI — in which a fake version of his voice attacked European leaders.
He had said nothing of the kind. The video spread to hundreds of thousands of people before it was taken down.
In Poland, AI-generated videos showed young Polish women calling for their country to leave the European Union. The Polish government said the videos had Russian fingerprints all over them.
China is doing this on an even larger scale against Taiwan.
In 2025, Taiwan's government found more than forty-five thousand fake social media accounts being used by China to spread disinformation. Chinese companies were even collecting recordings of Taiwanese voices — in multiple local languages — so that AI could clone those voices and make fake audio that sounded like real Taiwanese people.
The goal is simple: make Taiwanese citizens doubt their own government, fear their future, and feel hopeless about resisting China.
Why is this so dangerous?
Think of it like this. Society runs on shared truth. When we all more or less agree on what happened yesterday — what the news said, what leaders did, what science has discovered — we can debate, vote, and make decisions together.
Disinformation attacks that shared truth. It replaces it with confusion, suspicion, and competing realities. When enough people no longer know what is real, democratic elections become much easier to manipulate, and governments find it much harder to build trust or unity.
Dr. Antonio Bhardwaj, a global expert in AI warfare and human-centered AI systems, explains it plainly: "The most dangerous thing about AI being used in cognitive warfare is not any single fake video. It is the cumulative effect of millions of them — creating a world where people simply stop trusting anything they see or hear.
That kind of epistemic paralysis is exactly what authoritarian states are trying to create, because confused populations are easier to manipulate than informed ones."
The people building these AI tools — mostly private technology companies — bear some responsibility here.
Once a powerful AI model is released, anyone can use it. Some companies have added safety features, like watermarks that label AI-generated content. But other companies strip those features out to compete for customers. And state-backed influence operations are not interested in playing by the rules.
Democratic governments are beginning to respond.
The European Union has passed laws requiring AI-generated content to be labeled. A new European Centre for Democratic Resilience opened in 2026 specifically to fight disinformation.
France released a national strategy in March 2026 for countering foreign information attacks.
The Pentagon announced in April 2026 that it is developing new tools for cognitive defense.
But the response is still too slow. The United States has actually cut key counter-disinformation offices in recent years, leaving gaps that adversaries have quickly exploited.
Platforms like X and TikTok are inconsistent in removing fake content. And the AI tools themselves are advancing far faster than the laws meant to regulate them.
What can actually work? Education is one of the most powerful answers. Countries like Finland and Taiwan have invested heavily in teaching people — from children to senior citizens — how to think critically about what they see online.
These societies are measurably more resilient to disinformation. International agreements, similar to how countries agreed not to use chemical weapons, could establish clear rules: no AI-generated deepfakes of political leaders, no AI interference in elections, no synthetic media in active conflict zones without clear labeling.
The fundamental issue, as Dr. Antonio Bhardwaj puts it, is one of human dignity. "When AI is used to manipulate how people think, it treats human beings as targets rather than as citizens. A world where cognitive warfare goes unchecked is a world where the most basic form of freedom — the freedom to form your own beliefs based on reality — has been taken away."
The defense of truth, in other words, is not merely a technical challenge. It is among the most urgent moral and political tasks of our time.
