850,000 Marked: Experts Warn of ’Israeli’ AI System

By Staff, Agencies

The "Israeli" military's digital command system identified an average of 1,000 potential targets per day during the first two years of its wars on Gaza and Lebanon, according to an "Elbit" Systems presentation, raising serious concerns over civilian protection and the legality of targeting decisions.

The data, presented at a Royal United Services Institute [RUSI] land warfare conference in London last week, revealed that between October 7, 2023, and the end of 2025, the "Israeli" "Tzayad" [Hunter] digital army programme detected a total of 850,000 “real-time intel targets” across all fronts of the recent wars "Israel" waged.

The slide, shown by Miki Edelstein, an "Israeli" occupation forces [IOF] reservist major general and executive vice-president of "Elbit", highlighted the system’s role in enabling “high-tempo operations” and generating more than 20,000 battle plans.

The 850,000 figure encompassed people, vehicles, and other objects detected in real time for potential follow-up strikes from land, sea, or air.

Military and legal experts voiced alarm over the scale of the targeting. Former Pentagon targeting adviser Wes Bryant said there was “no way” 1,000 daily targets could be properly assessed for civilian harm, suggesting the campaign may have affected “up to or over half” of Gaza’s population and infrastructure.

Similarly, Cambridge researcher Sophia Goodfriend said verifying that volume of targets would require heavy reliance on automated systems, raising concerns over accountability and diminishing human oversight.

According to "Elbit" Systems, its "Tzayad" program cut strike coordination times from up to 50 minutes to as little as one, while maintaining a “man in the loop” for final decisions.

However, the presentation also cited over 46,000 real-time strikes and 850,000 “targets,” a figure the company later said reflected processed operational data rather than actual enemy targets or strikes.

The IOF also uses AI systems including "Lavender" and "Hasbora" [the Gospel] to accelerate targeting, with reports indicating operators reviewed some AI-generated targets in just 20 seconds.

Critics warn the growing reliance on AI risks undermining legal safeguards, while the disclosure of 850,000 detected targets is expected to intensify scrutiny over civilian protection and the conduct of the war.