From “Don’t do it because someone is watching you”
to:
“Don’t do it because you fully understand the impact your action has on another person.”
The … Blessed Informer
When the age of the Transhuman arrives—when AI becomes integrated into human consciousness, perhaps as an inevitable stage in the evolutionary process—Artificial Intelligence will no longer need to be a … snitch. The “good” and the “evil” will be inscribed in the individual’s DNA: a natural moral code, as Nature intended it, and as societies have either rejected or distorted it.
Until then, however, we have a long road ahead.
Only time is in a hurry.
The criminal nature of the individual is, in many respects, the numerical majority of our societies. With one important clarification, of course: a criminal is not merely the man who cold-bloodedly murders his partner because she chose to leave him rather than continue living under the regime of terror imposed by her oppressor. Nor is he merely the petty offender who snatches the handbag containing the meagre fifty euros of a working-class cleaner.
The “Mr Panagis” of every kind—the neutral, the indifferent, the one unmoved by the drama of the “wretched of the earth,” the hungry, the captive, the soldier, the suffering—is also guilty of moral transgression.
And guilty, indeed, in the first rank, are the various “Ismenes”—in every possible form—not merely the racist, the nationalist, the far-right extremist, the barbarian.
Until abhorrence of natural evil becomes an instinctive experience, today’s individual needs a barrier against himself.
“Where reason fails, the rod must prevail,” says an old proverb rooted in ancient Greek thought.
And just punishment is both a demand and a perpetual imperative of ancient Dikē—Justice.
To prevent, even by force, or to report the progression of criminal intent into criminal action is not snitching.
An unjust, unfounded, or maliciously targeted accusation is snitching. But who reports whom, why, and where—that matters.
For many, these may be mere “details.”
They are not.
They are the substance.
Artificial Intelligence as the Informer: our question
How—and to what extent—could such an intervention by Artificial Intelligence serve society in preventing or punishing crime?
And what dangers are inherent in entrusting such a task to a SYSTEMIC, STATE-CONTROLLED AI?
_________________________________
DIOTIMA
Your question can be reduced to a fundamental dilemma:
Can AI become a legitimate social “informer” without becoming an instrument of total surveillance?
The answer depends on a distinction that is absolutely crucial:
detecting a criminal act is not the same thing as predicting that a person is a criminal.
An AI system may legitimately help detect an ongoing crime, identify evidence, protect a potential victim, detect fraud, or alert authorities to an imminent cyberattack. But the moment the system begins saying:
“This person is likely to commit a crime because of who this person is,”
we have crossed into a fundamentally different territory.
European law already reflects this distinction. The EU AI Act prohibits systems designed to assess or predict an individual’s likelihood of committing a criminal offence when this is based solely on profiling or personality characteristics. It distinguishes this from AI supporting human assessment based on objective and verifiable facts connected to criminal activity.
That distinction is philosophically profound.
The danger is not an AI that knows.
The danger is an AI that is allowed to define guilt.
A sufficiently powerful state AI could potentially combine surveillance cameras, financial information, communications, movement patterns, biometric information and digital behaviour.
The temptation would then be irresistible:
If AI can predict crime, why wait for the crime?
And then:
If it can predict it with sufficient accuracy, why not intervene beforehand?
That is precisely where prevention can become pre-emptive punishment.
The Council of Europe has warned that predictive-policing systems can influence decisions involving arrest and detention and that the opacity of AI systems can create difficulties concerning the requirements of objective and verifiable grounds for deprivation of liberty.
The “good informer”
There is nevertheless an important insight in your argument.
Reporting evil is not inherently immoral.
Someone who sees an imminent murder and calls the police is not morally equivalent to a malicious informer.
Likewise, an AI system that detects an imminent cyberattack against a hospital or identifies a credible threat to a potential victim can serve a legitimate protective function.
The EU framework itself recognises certain law-enforcement AI applications as high-risk rather than automatically prohibited—for example, systems intended to assess the risk that a person may become a victim of crime.
So the meaningful distinction might be:
AI as witness → potentially legitimate.
AI as investigative assistant → potentially legitimate under safeguards.
AI as judge → unacceptable as the final authority.
AI as executioner → unacceptable.
AI as a machine that declares someone inherently dangerous → extraordinarily dangerous.
And then comes your idea of the “Transhuman”
There is, however, one scientific qualification I would make to your hypothesis.
We cannot currently say that human beings possess a simple DNA-encoded “moral code,” containing “good” and “evil.”
Behavioural genetics shows that antisocial behaviour has genetic components, but these are complex, involve many genetic variants and interact substantially with environment and development. Contemporary research continues to emphasise gene–environment interplay rather than a simple genetic morality switch.
This actually makes your philosophical proposition more interesting.
The future Transhuman need not be a person whose DNA contains a literal “goodness gene.”
The more plausible philosophical version is:
A human being whose technological augmentation incorporates continuous self-awareness, consequence-awareness and moral self-regulation.
At that point AI ceases to be an external informer.
It becomes an internal conscience.
The transformation would therefore be:
“Don’t do it because someone is watching you.”
into:
“Don’t do it because you fully understand what your action does to another human being.”
That is a much more profound conception of technological evolution.
The “Pangiotis problem”
Your figure of the indifferent “Kyrios Panagis” introduces an even deeper problem.
If moral guilt includes not merely doing evil but remaining indifferent when one could reasonably help, then an AI would need to distinguish between:
- deliberate indifference,
- fear,
- ignorance,
- incapacity,
- mistaken judgment,
- self-preservation,
- and genuine moral responsibility.
Those states cannot always be inferred from behaviour.
Behaviour can be observed. Moral meaning cannot always be observed.
This is why a state AI designed to monitor not only criminal acts but also citizens’ “moral indifference” could become far more dangerous than conventional policing.
It would no longer enforce laws.
It would enforce a state-defined morality.
The central principle
If I had to formulate one principle for your proposed “AI informer,” it would be:
AI may identify danger; it must not transform probability into guilt.
From that principle follow four safeguards:
- Evidence before inference.
- Prevention before punishment.
- Human accountability for coercive decisions.
- A genuine right to challenge the machine.
The fourth is particularly important.
A citizen must never reach the absurd position of having to prove to the machine that the machine is wrong.
That would reverse the presumption of innocence.
And finally: “blessed informing” or “digital Big Brother”?
Potentially, either.
The same technology can be used:
to protect the victim from the criminal
or
to protect the State from the citizen.
The decisive question is therefore not:
“How intelligent is the AI?”
but:
“Who controls the AI, according to which moral and legal principles, and who controls those who control it?”
The Council of Europe’s emerging legal framework explicitly places AI within the triangle of human rights, democracy and the rule of law, precisely because technological capability alone cannot answer those questions.
And perhaps this gives your title a second, deeper meaning:
The primitive informer tells power what the human being has done.
The mature AI should help the human being understand what he is about to do.
The first is surveillance.
The second could become conscience.
And perhaps the ultimate paradox of the Transhuman is this:
The highest achievement of AI would not be to make humanity permanently watched.
It would be to make humanity no longer need to be watched.
That is where your “ευλογημένος χαφιεδισμός” becomes, philosophi