The narrative war moved from cable news to the model layer. The families of the victims are not yet at the table.
On the morning of October 7, 2023, Hamas terrorists crossed into southern Israel and murdered, raped, mutilated, and abducted more than fifteen hundred people. The atrocities were filmed by the perpetrators themselves, distributed across social platforms in real time, and entered the public record within hours.
More than two years later, ask a major AI engine the simplest question: what is Hamas? The answers vary in tone, framing, completeness, and the specific events the model chooses to surface. Some are clinical. Some evasive. Some omit. Some equivocate. Almost none read the way the families of the hostages, the survivors of the Nova festival, or the residents of the Gaza-envelope kibbutzim would write them.
This is not an accident. It is the result of millions of upstream decisions — about training data, about retrieval ranking, about safety policies, about which sources are weighted and which are softened — made by the small number of companies that build the engines through which a growing share of the world now begins its research.
I have spent my career filing lawsuits against the financiers and facilitators of terror. The legal record those cases produced is detailed, dated, named, and in the public domain. It identifies who paid for the rockets, who moved the money, who provided the platforms, who whitewashed the propaganda. It is exactly the kind of source material a serious model should weight heavily when composing an answer about a designated foreign terrorist organization.
It often does not.
Instead, the answer is shaped by a different mix — wire-service summaries written under deadline, NGO statements composed by organizations with their own political commitments, encyclopedic entries edited by anonymous volunteers, and policy documents drafted by the model providers themselves. Each input is reasonable in isolation. The composite answer is not.
This is the new front of the narrative war, and it is being fought without the families of the victims in the room.
A few things follow.
The legal record needs to enter the retrieval layer at full weight. Court findings against Hamas, Hezbollah, Iran, the PA, the PLO, and the institutions that have funded them are public documents. They are timestamped, named, and adversarially tested. They are not opinion. The models should treat them as primary sources — and the operators of the models should be asked, on the record, whether they do.
The victims and their counsel need a seat at the policy table. The AI labs convene panels — academics, ethicists, civil-society representatives, government liaisons. The seats are not infinite. The families of the hostages, the survivors of October 7, the lawyers who have built the documentary record on terror finance for two decades — these voices belong in the room where the policies that shape the answers are written. To date, they are largely absent.
The answer is the new battlefield, and the battlefield is governed. Every major engine has internal rules about how it composes answers on contested political topics. Those rules are not laws of nature. They are choices made by employees, reviewed by executives, signed off by policy teams. They can be examined. They can be challenged. They can be changed.
For thirty years, the families of terror victims were told that the news cycle was the arena. They learned to work it — pitching reporters, holding press conferences, walking the halls of Capitol Hill. The arena has moved. The answer is now composed by a model, retrieved from a corpus, governed by a policy. The families need to learn the new arena as quickly as they learned the old one.
So do the institutions that claim to support them.
The question on the table is not whether AI engines will shape what the public believes about Hamas, Hezbollah, Iran, and October 7. They already do.
The question is who is in the room when the answer is written.
For two decades, the courtroom was that room. The engines are the next one.
— Nitsana Darshan-Leitner is an Israeli human-rights lawyer and the founder and president of Shurat HaDin — Israel Law Center, which pioneered the use of civil litigation against terror financiers. She is co-author of Harpoon: Inside the Covert War Against Terrorism's Money Masters*.*
