When AI Ethics Becomes a New Egypt
The Vatican is right to enter the debate on artificial intelligence. But it should not be the only religious voice standing at the threshold.
Pope Leo XIV’s forthcoming encyclical Magnifica humanitas, devoted to safeguarding the human person in the age of artificial intelligence, marks an important moment. It places AI where it belongs: not merely among technical innovations, market opportunities, or national-security instruments, but inside one of the oldest questions of civilization: what may power do to a human being?
The staging itself matters. This is not only a document being released. It is a scene being assembled. A pope, cardinals, theologians, and Christopher Olah of Anthropic appear together around the same problem: how advanced AI is to be morally named, governed, explained, trusted, and limited. Before the world has even learned how to contest these systems properly, a moral theater of technology is being formed.
That is precisely why another voice is needed.
Not an anti-Catholic voice. Not a secular sneer. Not another predictable argument about “religion and technology.” What is needed is a Hebrew interruption.
Christian language, especially in its modern Catholic social form, tends to speak of the dignity of the human person, humanity, universal responsibility, and the common good. This is not weak language. It has protected workers, the poor, the vulnerable, and the wounded against many forms of instrumental power. In the age of AI, it can still serve as a barrier against reducing the human being to data, function, labor, risk, or target.
But the Hebrew tradition does not begin from the universal human person.
It begins from commandment, interruption, prohibition, witness, the stranger, the widow, the orphan, the debtor, the worker, the defeated, and the enemy. It does not first ask whether the human being possesses dignity within a universal moral order. It asks whether an apparatus has begun to behave like an idol.
This distinction matters.
An idol is not only a statue. An idol is any human-made structure that receives the authority to decide reality while hiding the hands that made it. An idol does not need incense. It needs obedience. It needs people to say: the system knows. The model assessed. The algorithm predicted. The platform detected. The risk score indicated. The target was generated. The decision was procedurally valid.
At that moment, the machine has not merely assisted judgment. It has absorbed responsibility.
A Hebrew response to AI ethics must therefore begin with a prohibition: no model may become an idol of decision.
This does not mean that AI must be rejected. Judaism is not a romance of technological innocence. The Hebrew Bible is full of tools, measurements, architectures, laws, instruments, records, contracts, and systems of transmission. The question is never simply whether a tool exists. The question is whether the tool abolishes responsibility before the face of the other.
Here the Hebrew tradition offers something sharper than generic humanism.
The stranger, the widow, the orphan, the poor, the debtor, the worker, the slave, the accused, the defeated, and the enemy are not decorative moral symbols. They are tests of power. They reveal whether a society’s order has become an apparatus of disappearance.
AI will not first become dangerous when it produces killer robots. That is the spectacular case, and it deserves condemnation. But the deeper danger begins earlier. It begins when a person becomes administratively invisible, when poverty becomes a probability, when migration becomes a risk profile, when employment becomes behavioral prediction, when policing becomes statistical anticipation, when war becomes pattern recognition, when the face of the other is replaced by a confidence score.
The Hebrew question is therefore not only: does AI respect human dignity?
The Hebrew question is: who has been made removable?
This is where a Hebrew voice that is not captured by state sovereignty becomes necessary. This does not mean indifference to Jewish vulnerability. It does not mean a cheap rejection of Israel’s history, fear, or need for protection. It means something more precise: the refusal to identify Hebrew responsibility with the sovereignty of any state.
Once the state becomes the highest interpreter of Jewish existence, the prophetic wound is sealed. The state begins to speak in the name of survival, and survival begins to excuse every apparatus built in its name.
That danger is not theoretical. In every modern state, including Israel, AI can become a technology of preemptive suspicion. It can organize who is seen, who is classified, who is followed, who is delayed, who is struck, who is forgotten, and who is never allowed to appear as a full human presence in the first place.
A Hebrew ethics of AI must say clearly: no people’s trauma gives it the right to automate the reduction of another people to risk.
This sentence should not be read as a slogan. It is a theological and political limit.
The Exodus story is not only a memory of liberation. It is also a permanent warning against becoming Egypt. Egypt is not merely a place. It is a system. It counts bodies, extracts labor, manages fear, stores grain, builds monuments, and turns human beings into production material. Pharaoh does not need to hate every slave personally. He only needs the bricks.
Modern Pharaoh does not always speak with cruelty. He often speaks with efficiency. He says: optimize. Predict. Secure. Prevent. Manage. Scale.
That is why AI ethics without Exodus becomes dangerous. It can become a manual for humane brick-making.
Catholic social teaching rightly remembers the worker. The analogy with Rerum novarum is powerful, because AI may become to the twenty-first century what industrial capitalism was to the nineteenth: a force that reorganizes labor, value, dependence, and social power. But a Hebrew response must also remember Egypt.
The issue is not only whether AI will displace labor or violate workers’ rights. The issue is whether work itself will be transformed into a managed condition of dependency, where the worker is no longer exploited only by an owner, but shaped by prediction, ranking, scheduling, surveillance, and behavioral correction. The Hebrew distinction is not simply between labor and capital. It is between service and bondage.
That distinction is decisive.
A person may work, serve, build, repair, create, and contribute. But when the system begins to govern the worker’s future through invisible scoring, behavioral nudging, predictive discipline, and automated correction, work approaches bondage in a new form. The whip has become statistical. The overseer has become procedural. The brick quota has become a dashboard.
This is why a Hebrew response cannot simply become another chapter in universal ethics. The Hebrew tradition does not dissolve the stranger into “humanity.” It does not allow the poor to disappear into “the vulnerable.” It does not convert the widow and the orphan into abstract moral subjects. It names them because power always hides its violence by generalizing its victims.
Universalism can protect.
But it can also blur.
A Hebrew interruption must keep the wound particular. It must ask not only what happens to “humanity,” but what happens to this stranger, this child, this worker, this refugee, this enemy, this person whose future has been closed by a system that claims only to predict.
The Vatican conference speaks of preserving human faces and voices. A Hebrew response should welcome that concern, but deepen it. In the Hebrew tradition, voice is not merely expression. It is command, witness, summons, interruption. The problem with AI is not only that it can imitate the human voice. The deeper problem is that it can produce voices without obligation, speech without witness, and decisions without anyone who can answer for them.
This is also where interpretability, however necessary, reaches its limit.
To interpret a model is not yet to answer for a decision. A system may become more transparent and still remain morally evasive if no one can be held responsible for what its classifications do to the stranger, the worker, the refugee, the accused, or the enemy. A Hebrew ethics of AI must therefore move beyond explainability toward answerability.
A rabbinic or Hebrew counter-voice should insist on four principles.
First, the prohibition of algorithmic idolatry. No model may be treated as a source of final reality. Its outputs must remain contestable, situated, interruptible, and answerable to human witnesses.
Second, the principle of the accountable witness. “Human in the loop” is not enough. A human rubber stamp inside a machine process is not responsibility. There must be someone who can be asked: what did you see, whom did you trust, what did you ignore, whom did you hand over to the apparatus?
Third, the protection of the stranger from predictive closure. A person may not be imprisoned inside the statistical shadow of their past. In Hebrew terms, this is the technological defense of teshuvah. A human being must remain capable of becoming otherwise. Any system that says “we know what you will be because we know what you have been” attacks the possibility of return.
Fourth, the Sabbath limit. There must be zones of life that are not optimized, scored, accelerated, ranked, extracted, or converted into predictive material. Sabbath is not leisure. It is the interruption of Pharaoh’s world. It says that creation itself requires a limit to production.
This is the Hebrew contribution the AI debate needs.
A Hebrew response should not enter this debate as one more religious contribution to a global ethics panel. That would already weaken it. Its task is not to add a Jewish value to a universal vocabulary. Its task is to ask what every universal vocabulary risks concealing: who has been counted before being heard, classified before being seen, and removed before becoming a neighbor.
The Pope can rightly speak of dignity. A Hebrew voice must ask when the apparatus becomes an idol, when the stranger becomes removable, and when responsibility disappears into procedure.
These are not simply rival religious vocabularies. They are different thresholds. Dignity protects the human person against degradation. The Hebrew interruption protects the world against the apparatus that makes degradation procedurally invisible.
The question before us is not whether artificial intelligence will serve humanity. That phrase is too easy. Every empire has claimed to serve humanity in its own language.
The question is whether AI will become a new Egypt without a visible Pharaoh: a system in which no one is guilty because everything was only a procedure.
That is where the Hebrew voice must stand.
Not outside technology.
At the gate.
And at the gate it must say: no machine, no state, no company, no priesthood of engineers, and no frightened nation may erase the stranger in the name of order.
Yochanan Schimmelpfennig
