Article

A Prediction Must Never Become an Unanswerable Accusation

16 November 2026

Being told that you are a danger is difficult enough. Being unable to discover why, or what would persuade anyone otherwise, would leave you in a peculiar kind of captivity. You might remain free to walk down the street while losing any meaningful way to answer the suspicion following you.

That is the human problem worth keeping in view when we discuss artificial intelligence in terrorism prevention. A prediction may justify further investigation. It cannot fairly become an accusation that the person concerned has no means of challenging.

Financial News (파이낸셜뉴스) reports that a Korean National Police University study, published in an international journal, examines human rights risks associated with AI-assisted terrorism prevention. The available reporting does not establish the study’s methods, which systems it considered or whether it documented particular harms. It would be wrong to supply those details by imagination. The reported concern nevertheless opens a substantial ethical problem.

Preventing terrorism means acting before an intended harm becomes an accomplished one. Waiting for certainty can carry terrible consequences. There is a serious argument for using tools that might help people recognise danger earlier, and any discussion of rights that ignores those who could be attacked is incomplete.

But urgency does not settle what a prediction entitles an authority to do.

Suppose a system identifies someone for closer attention. The important questions then concern the people making decisions: what evidence will they require, how far may they intervene, and what could make them change their minds? A human signature at the bottom of a decision offers little protection if the person signing has neither the knowledge nor the practical freedom to disagree.

There will be information that cannot safely be disclosed publicly. That makes independent scrutiny more important, although it does not automatically resolve the difficulty for someone trying to contest a decision whose basis remains hidden.

Living together requires us to accept some uncertainty about one another. We cannot make every person prove in advance that they will never cause harm. If we pursue that ambition through increasingly powerful prediction, ordinary life risks becoming a continuing examination in which nobody knows what counts as passing.

The practical burden should remain with whoever restricts another person’s life: explain the grounds, permit meaningful challenge, and reconsider the decision. An authority should not acquire greater freedom to act merely because its reasons have become harder to understand.