Israel's military and its Shin Bet security service acknowledged on Saturday, for the first time, that a dedicated joint unit has spent nearly three years identifying, locating and killing every person assessed to have taken part in the October 7, 2023 attack. The unit is called Nili, and the acknowledgment came with something rarer than the admission itself: the state's own published accounting of the program's results [1].
By the unit's own figures, roughly 5,000 people participated in or assisted the October 7 attack. Of those, 2,800 people on the target list have been killed. That total includes 1,200 who died during the initial assault itself, which means the manhunt conducted since accounts for roughly 1,600 killings. A further 300 alleged perpetrators are in Israeli custody awaiting trial before military tribunals [1]. Every one of those numbers is the IDF's own. None has been independently verified, and this piece reports them as the military's claims, not as established fact.
The arithmetic the ledger implies is worth stating plainly. If 5,000 people participated or assisted, and 2,800 are dead and 300 detained, then roughly 1,900 people on the unit's assessment remain neither killed nor in custody. The published accounting is, among other things, a statement that the program is not finished.
The methods are a catalog of the modern surveillance state applied to a manhunt. According to the acknowledgment, Nili works from facial recognition, AI-assisted analysis of imagery, human intelligence sources, interrogations of detainees, and footage the attackers recorded themselves on GoPro cameras during the assault [1]. The attackers' own documentation of October 7, in other words, became the evidentiary spine of the campaign to kill them.
An IDF lieutenant identified only as "Lammed" described what an operation feels like from inside the unit: "You feel the tension [when an elimination operation is underway]. You feel the quiet, the seconds before the missile hits." [1]
What makes this a record worth setting down is not that Israel hunted the perpetrators of October 7. That was assumed, stated in general terms by Israeli officials since the first weeks of the war, and surprising to no one. What is new is the formal acknowledgment of a standing unit with an explicit mandate to kill a defined list of people, published together with the list's running totals. States have run targeted-killing programs before. States have rarely, if ever, issued their own audited-sounding ledger of one.
The acknowledgment also puts two paths side by side in the state's own numbers. Three hundred alleged perpetrators are awaiting trial before military tribunals, which establishes that a judicial route exists and that Israel considers its evidence against at least some participants strong enough to present in a courtroom. The roughly 1,600 people killed in the manhunt since October 7 will never enter that courtroom. The published figures do not explain, case by case, why one alleged participant was detained for trial and another was killed, and the acknowledgment offers no criteria distinguishing the two groups. That silence is itself part of the record: the state has published how many it killed and how many it will try, without publishing the rule that sorted them.
The timing and the fact of the acknowledgment serve identifiable purposes. Publicly confirming a unit that will hunt every participant is a deterrence message aimed at anyone weighing future attacks. Domestically, it is an accountability statement to an Israeli public that has demanded answers for the roughly 1,200 people killed on October 7, delivered in the form of a body count. Both purposes are served by publication, which is presumably why a program that ran in secret for nearly three years is now official.
What the record now contains, in the state's own voice, is this: a list of roughly 5,000 names, a count of 2,800 dead against it, 300 awaiting trial, and a standing unit assigned to the remainder [1].
