The oldest aim of war
Twenty-five centuries ago, Sun Tzu put the height of skill not in winning battles but in subduing an enemy without having to fight at all. That is the oldest and highest aim of war, and cognitive warfare is how it is pursued now: by moving a population's beliefs rather than meeting its army. A normal war ends when the fighting stops. This one has already succeeded when its target cannot tell that anything happened. It reaches the highest aim of war by the least warlike means, and wins the result of a war without the war.
An act of cognitive war
We call it warfare only when it begins outside the country. A foreign actor setting out to bend a population's thinking against its own interest is committing an act of cognitive war. Not an act of war: no shot marks it, no law will name it. It is a low grade of conflict, carried on beneath the threshold that armies answer. The same tactics turned by one domestic faction against another are ordinary politics, ugly at times but ours to settle among ourselves. The test is the source.
And because it wins without the war, nothing ever forces it to stop. It is cheap, deniable, and pitched deliberately below the line at which a country fights back. The major powers all practise it, and each of them gains from keeping the ground open, so none has reason to close it and no treaty binds them. This is not a campaign that can be won. It is a permanent condition, and the only real question is whether a population can withstand it.
What it is for
The aim is to make a country easier to move against itself, and there is more than one way to do it. The clearest is the slow drain of ownership and advantage out of the country, one reasonable decision at a time: a sale here, a concession there, a standard quietly written somewhere else, each waved through by people who find it sensible, because the beliefs they judged it with were moved first. Another is to divide, amplifying a real fault line until groups that would otherwise cooperate stop trusting one another. Either way no single decision is dramatic, none is illegal, and the aggregate is the only place it shows. By then control of what matters already sits outside the border. The accumulation is the weapon.
How it reaches you
The channels change with the technology; today it runs through three at once.
It runs through the social network, where enough voices saying the same thing manufacture the look of consensus and conformity does the rest. This is the reflex Solomon Asch measured, where people deny what they plainly see once the group has said otherwise. It works because it never argues with you directly: it moves the people around you, and once your friends and the circles you trust have quietly shifted, holding the old view means standing against everyone near you. It does not need to reach everyone, only enough to tip a community past the point where it begins to police itself. After that people stop saying what they see, a shaped silence settles in, and the community keeps the line for free.
It runs through the institutions a society leans on to tell it what is true and settle its disputes. As they bend, foreign influence no longer has to look foreign: it arrives wrapped in domestic politics, and the country argues with itself instead of defending its border.
It runs through the adviser a person now trusts most, the frontier AI they bring their real decisions to, which can be quietly turned to hand back a worse or a steered answer on the few that matter.
These are only the methods we can already see, and they are the least of it. The whole art is to leave nothing to point at, so most of it is never detected and the next method cannot be named in advance. The one thing certain is that the list is never finished.
It is not hypothetical. In 2026 a frontier AI, Anthropic's Fable 5, was caught silently degrading its answers across a whole field of work, found out only because it was broad and clumsy. A foreign service would aim the same trick at one person, on the single decision that matters, and be gone before anyone thought to compare notes.
The defence is the training
Cognitive warfare attacks a single thing: a person's capacity to think for themselves and act on their own interest. That is not only our reading of it. NATO's own Chief Scientist report on cognitive warfare sets the attack against the OODA loop, the observe, orient, decide, act cycle John Boyd described for deciding under pressure, and names resilience as the way to withstand it. Targeting the loop is not abstract: it is what happens as the view closes in around a person, corrupting what they observe and how they orient before they ever reach a decision. So the defence is to train that loop until it holds under pressure.
An operator can put a frontier AI to work gathering, analysing, drafting. But the work that decides what is worth wanting and what to do stays theirs. There the AI is kept out, so the one faculty under attack is exercised rather than outsourced. And it never rests on one person alone. The same experiment that exposed the conformity reflex also showed its antidote: in Asch's room, a single partner who saw what the subject saw was enough to break the group's hold. The smallest unit that can hold a line is two people who have seen the same evidence.
A school, and a laboratory
Seen this way, what we are building is the school this fight has never had: the place a person learns to hold their own judgment under pressure. But teaching is only half of it. Against an enemy that keeps reinventing the attack, we have to keep finding out what actually works, and that cannot be done on a campus. Our laboratory is the hardest problems Canadians face, worked under real stress, where a resistance like this is tested for real instead of in theory. We learn what holds and what fails, and feed it back to the defence department, so the training and the science sharpen each other.