Israel has employed artificial intelligence on a scale never seen before in its military campaign in Gaza, with many of the systems created through partnerships between active-duty intelligence soldiers and reservists employed at major tech companies, The New York Times reported Friday, citing sources from European, American, and Israeli defense circles.
Although Israel’s integration of AI into military operations was already known, the latest revelations outline a broader range of tools, including a chatbot trained across multiple Arabic dialects to assess public opinion, an AI-enhanced audio detection system that identifies targets based on sounds like bombings and airstrikes, and a facial recognition technology capable of matching obscured or injured individuals to known identities.
Israeli officers who spoke to the Times said that AI has been useful in locating hostages and streamlining operations but admitted that the systems can sometimes fail. According to the Times, “some officials have struggled with the ethical implications of the AI tools” because of concerns over surveillance expansion, the risk of civilian casualties, and mistaken arrests.
Responding to a Times inquiry, the IDF said it could not comment on the specifics of the technology in use, but stressed that Israel “is committed to the lawful and responsible use of data technology tools.”
According to four Israeli officials cited by the Times, the military rapidly approved the use of these AI systems following the Hamas-led assault on October 7, 2023, when thousands of terrorists attacked southern Israel, murdering roughly 1,200 people and abducting 251 others.
Some AI applications previously reported include a virtual reality platform designed to assist soldiers navigating urban combat zones. The Arabic-language chatbot was first brought to light by the Israeli-Palestinian outlet +972, which also reported on a system called Lavender that allegedly helped the IDF compile a list of 37,000 human targets based on connections to Hamas. The military has denied relying on AI to create any such kill list.
The majority of these AI capabilities were reportedly developed in an innovation center under the Military Intelligence Directorate’s Unit 8200, known as “The Studio,” which, according to the Times, linked serving soldiers with reservists employed by firms like Meta, Google, and Microsoft.
Avi Hasson, CEO of Startup Nation Central, which fosters ties between investors and Israeli startups, told the Times that the reservists “brought know-how and access to key technologies that weren’t available in the military.”
Meta and Microsoft chose not to comment on the matter, while Google confirmed that it employs reservists in many countries but said their military service activities were unrelated to the company’s work, according to the Times.
Citing three Israeli and American sources familiar with the matter, the Times reported that the IDF incorporated AI into older surveillance systems in October 2023 to track Ibrahim Biari, the commander of Hamas’s Central Jabalia Battalion.
The enhanced system reportedly enabled Israeli forces to monitor Biari’s communications at a Hamas tunnel network beneath the densely populated Jabalia refugee camp. A strike was approved despite internal warnings that several residential buildings would need to be hit to ensure Biari’s elimination. The resulting attack, which allegedly caused dozens of civilian deaths, is under military investigation. The IDF told the Times it “was unable to provide any further information until the investigation is complete.”
The audio surveillance tool used to locate Biari has since been refined to more accurately pinpoint individuals, and it has reportedly helped the IDF in efforts to locate hostages, Israeli officers told the Times.
Following the October 7 attacks, Israel set up temporary checkpoints between northern and southern Gaza, equipped with cameras capable of sending high-definition images to an AI facial recognition system, the Times reported.
However, two Israeli intelligence officers told the Times that the technology occasionally struggled to recognize obscured faces, resulting in the mistaken detention and questioning of innocent Palestinians.
According to three Israeli officers, “The Studio” also produced a large language model trained to handle spoken Arabic across various dialects — a challenge since most training materials are in classical Arabic, while the IDF had access to decades’ worth of intercepted texts, phone transcripts, and social media content written in spoken dialects.
The Times reported that the AI language model was developed during the early months of the war and became part of the military’s multimedia analysis platforms, enabling complex searches across image and video databases.
Despite its capabilities, the chatbot had issues with modern slang and English transliterations, at times producing incorrect results, such as “returning photos of pipes instead of guns,” according to the Times. Intelligence officers fluent in the dialects had to review the output manually.
Even so, two Israeli intelligence officials cited by the Times said the chatbot dramatically improved the speed of intelligence gathering and analysis.
Three Israeli officers said the technology was instrumental in assessing whether public sentiment in the Arab world would demand retaliation after Israel’s September assassination of Hezbollah leader Hassan Nasrallah.
{Matzav.com Israel}