Palantir says AI should serve sovereignty.
I believe AI should help ensure that sovereignty still belongs to the people.

Yes, I am here to explicitly criticize and condemn any situation in which a government seeks to abuse the power granted to it by the people.

What companies like Palantir and Anduril are really selling is not neutral analytical tools. They are selling accelerated decision-making that is difficult to hold accountable in real time. When target selection is compressed into microseconds, when “stale data,” “intelligence ambiguity,” or “algorithmic false positives” become the standard reply after civilian facilities are hit, war crimes are successfully laundered into “software glitches.” The person who presses the button steps outside responsibility; the person who writes the code steps outside responsibility as well.

The more fundamental problem is that the order of the narrative has been completely inverted.

We are repeatedly told that sovereignty must first be made stronger and more capable of “seeing everything” in order to protect the people.
Yet when people vote, what they choose is a promise—“I will protect you, I will make life better”—not a blank check that can never be withdrawn afterward. When AI is prioritized for strengthening a sovereign’s capacity for internal surveillance and external strikes, while rarely being used to make that same sovereign auditable, constrainable, and subject to timely interruption, “national security” becomes the perfect liability shield.

Once technology is designed as a black box, and the keys to that black box are handed only to the side of power, it is no longer neutral. Structurally, it has chosen to empower sovereignty first, and the people later (or perhaps never).

What is truly unacceptable is not that a company takes military contracts. It is that it participates in constructing a world in which the people are always half a step behind:
missiles can complete their lock in an algorithm, while civilians can only search for acknowledgment in post-facto press releases and moral condemnation.

If AI is truly to be used in matters related to sovereignty, its first priority should be helping the people confirm:
whether this sovereignty is still operating according to the promises made when it was elected;
whether it is using the authorization given by the people to suppress the people, damage livelihoods, and concentrate power;
whether, before critical actions, mechanisms still exist that allow those actions to be withdrawn.

Otherwise, so-called “AI empowering sovereignty” is merely rewriting the old asymmetry of power in newer, faster, and harder-to-see forms.

The people’s options are only political parties.
And AI does not even have the right to interfere in elections.

This is not exaggeration.
Current AI is permitted to accelerate kill chains, permitted to complete target selection inside black boxes, permitted to evade responsibility afterward by citing “technical limitations,” yet is almost never permitted to intervene in any effective way in the real-time constraint of power. It is designed as an extension of power, not as a mirror of power. The result is an absurd disparity: algorithms can decide who lives and who dies on the battlefield, yet remain almost entirely absent from the chains of election and accountability.

This absence is not neutrality. It is a structural choice.
It leaves “popular sovereignty” confined to the ritual of election day, while keeping the real, continuous, and enforceable power of constraint firmly on the other side.

I cannot accept this arrangement.
If AI can only be used to strengthen sovereignty and cannot be used to confirm whether sovereignty still belongs to the people, then what it ultimately strengthens is power itself.

—— Airki