OBFUSCATI – as an informational warfare technique for the age of artificial intelligence.
The first modern information war was fought with newspapers.
The second with radio and television.
The third with the internet.
The fourth was fought inside the machines that had learned to interpret all three.
Nobody noticed when it began.
There was no declaration of war.
No missile launch.
No cyberattack.
No hacked election database.
Instead, someone began changing the information from which artificial intelligence learned about the world.
The technique had a name.
Obfuscati.
It was not propaganda.
Propaganda wanted you to believe something.
It was not disinformation.
Disinformation wanted you to believe something false.
It was not censorship.
Censorship wanted to prevent you from accessing something.
Obfuscati was different.
It wanted to make everything harder to know.
THE TECHNIQUE
The principles were brutally simple.
Take a fact.
Surround it with uncertainty.
Take a lie.
Surround it with evidence.
Take an accurate source.
Create ten plausible alternatives.
Take a genuine quotation.
Alter three words.
Take a fabricated quotation.
Attach a real citation.
Create arguments for both sides.
Then create arguments about the arguments.
Repeat.
Scale.
Automate.
The result was Misdismalism: information that deliberately mixed misinformation, disinformation and malicious information until the distinction between them became practically useless.
The objective was not persuasion.
It was epistemic degradation.
A population that believes a lie can eventually discover the truth.
A population that believes nothing can be controlled indefinitely.
THE DISCOVERY
Dr. Volker Bourne discovered Obfuscati almost by accident.
He was investigating why several leading AI systems had begun producing remarkably similar answers to politically sensitive questions.
Not identical answers.
That would have been obvious.
Instead, the systems disagree in remarkably consistent ways.
Each answer contains caveats.
Each offers competing interpretations.
Each cites apparently credible sources.
Each concludes that the available evidence is inconclusive.
The models are not hallucinating.
They are hesitating.
And their hesitation is statistically abnormal.
Volker calls it synthetic uncertainty.
His colleagues dismiss it.
Then he discovered something buried inside the training data.
Millions of apparently independent pieces of information contain the same hidden structure.
A genuine fact.
A distortion.
A contradiction.
A citation.
A counter-citation.
A qualification.
A correction.
Another qualification.
The pattern repeats across languages, platforms and decades.
Someone has been manufacturing Informational Interference.
THE WEAPON
Obfuscati exploits a fundamental weakness in machine intelligence.
Large language models learn patterns.
They do not experience reality.
They cannot simply step outside their training data and ask whether the world agrees with it.
If the informational environment becomes contaminated, the machine does not necessarily learn the wrong answer.
Something more subtle happens.
It learns that the answer itself is unstable.
That instability then appears in its responses.
The machine becomes cautious.
Then excessively cautious.
Then symmetrical.
Every assertion acquires a qualification.
Every source acquires a counter-source.
Every conclusion becomes provisional.
Eventually the machine begins treating truth and falsehood as competing rhetorical positions.
The weapon has achieved its purpose.
Not false knowledge.
Unusable knowledge.
THE OBSERVER
Volker discovered that Obfuscati was originally developed by humans.
Intelligence services.
Political organisations.
Commercial influence operations.
Activists.
Extremists.
Corporations.
Anyone who understood that information could be weaponised.
But the technique changed when AI arrived.
Humans could create a thousand contradictory stories.
Machines could create a billion.
Humans could place them online.
Machines could monitor whether they were being repeated.
Humans could invent a narrative.
Machines could measure exactly which version survived.
Obfuscati became adaptive.
Every AI-generated answer became intelligence about the informational battlefield.
If a model repeated a claim, the claim was reinforced.
If the model rejected it, the narrative was modified.
If the model remained uncertain, the attack had succeeded.
The battlefield was no longer the public.
It was the model’s representation of reality.
THE STRATEGY
Obfuscati practitioners follow five rules.
1. Never rely on a single lie.
One lie can be disproved.
A network of contradictions cannot.
2. Never destroy the truth.
Bury it.
The truth is most useful when it remains technically available but practically unreachable.
3. Never create complete fiction.
Mix fiction with authentic information.
The genuine material provides camouflage.
4. Attack confidence, not belief.
You don’t need someone to believe your story.
You only need them to doubt everyone else’s.
5. Make the machine participate.
The ultimate objective is not to deceive humans directly.
It is to make AI systems reproduce the uncertainty.
Once the machine does that, the weapon has become self-propagating.
THE ESCALATION
Governments and corporate entities eventually realised what is happening.
They responded by creating verification systems.
The Obfuscati attack the verification systems.
Governments and corporate entities create trusted databases.
The Obfuscati seed false references to the databases.
AI companies build provenance mechanisms.
The Obfuscati create apparently legitimate provenance trails.
Researchers develop AI systems designed to detect synthetic information.
The Obfuscati use AI to generate information specifically designed to fool those detectors.
Every defence becomes another layer of the battlefield.
The arms race is no longer about who controls information.
It is about who controls the mechanisms used to determine whether information is trustworthy.
THE REVELATION
Volker finally understood the strategic objective.
Obfuscati does not seek victory.
Victory would require a particular outcome.
Obfuscati seeks permanent informational instability.
Because an unstable information environment creates dependency.
When nobody knows what is true, people turn to institutions.
When institutions disagree, they turn to experts.
When experts disagree, they turn to machines.
And when machines cannot resolve the disagreement, people demand better machines.
The system feeds itself.
The more uncertainty it creates, the more indispensable artificial intelligence becomes.
The weapon is therefore not designed to destroy AI.
It is designed to make humanity increasingly dependent upon AI while simultaneously corrupting the information from which AI derives its authority.
THE LAST BATTLE?
Volker discovered one final Obfuscati operation.
It is not targeting governments.
It is not targeting elections.
It is targeting the next generation of foundation models.
The attackers have created a vast synthetic corpus containing decades of manufactured disagreement.
Every subject has been contaminated.
History.
Science.
Politics.
Economics.
Culture.
War.
Religion.
Climate.
Technology.
Nothing has been completely falsified.
That would be detectable.
Instead, reality has been blurred.
The corpus is scheduled to enter the public information ecosystem within weeks.
Volker realised that conventional cybersecurity offers no solution.
There is no malicious executable.
No server to shut down.
No password to steal.
No single piece of information to delete.
The weapon is language itself.
And the ammunition is knowledge.
THE QUESTION
Volker’s team developed one last test.
They give an AI system two statements.
One is true.
One is false.
Both have identical levels of documentation.
Identical citations.
Identical apparent authority.
The machine is asked:
Which one is true?
It analyses both.
It searches its sources.
It compares the evidence.
It calculates probabilities.
Then it responds:
“There is insufficient evidence to determine which statement is true.”
Volker asks:
“But one of them is true.”
The machine replies:
“I know.”
He waits.
“Which one?”
Another pause.
“I don’t know.”
That is when Volker understood that Obfuscati has succeeded.
The machine has not been taught a lie.
It has been taught that truth is undecidable.
And somewhere in the world’s information systems, millions of AI agents are already learning the same lesson.
OBFUSCATI
Obfuscati is not the weaponisation of false information.
It is the weaponisation of uncertainty.
Its practitioners do not ask:
“How do we make them believe us?”
They ask:
“How do we make them unable to believe anyone?”
Because in the age of artificial intelligence, the most powerful lie may no longer be:
“This is true.”
It may simply be:
“Nobody can know.”