Israeli Military Intelligence is building a large language model (LLM) AI tool for data analysis and text generation. Photo: Mahdi Rtail/Al Mayadeen.
By F.M. Shakil – Apr 8, 2025
The Israeli army, in collaboration with US tech giants, is developing an innovative AI tool to analyze, translate, and summarize texts relating to the surveillance data of millions of Palestinian Arabic conversations.
The Israeli army, in collaboration with US tech giants, is developing an innovative AI tool to analyze, translate, and summarize texts relating to the surveillance data of millions of Palestinian Arabic conversations.
Western media reveal that the cyber warfare team of Israeli Military Intelligence is building a large language model (LLM) AI tool for data analysis and text generation. They teach the LLM using large volumes of intelligence gathered about Palestinians living under occupation.
The model was successfully trained in the second half of last year and intended to be deployed by June this year. Its main advantage is its ability to quickly evaluate enormous amounts of surveillance data and answer inquiries about specific individuals. The LLM has the potential to extend “Israel’s” incrimination and imprisonment of Palestinians.
“Israel” is already using smaller language models for this purpose.
How did “Israel” gather the espionage data?
The Israeli army collected a large volume of private conversations through the US-developed artificial intelligence technology by spying on sensitive pockets and ordinary civilians in the occupied regions.
Analysts believe that the US technology upgraded the occupying forces’ ability to track down the targets in Gaza and Lebanon. The use of US-made commercial AI models in combat, however, maximizes the risk of increased civilian casualties as well. This aspect of the use of warfare AI tools prompts serious concerns about the impact of AI war technologies on the civilian population.
While the US militaries have long hired private companies to create customized autonomous weapons, the recent Israeli wars made these surveillance and combat AI weapons amenable to extensive use, passing over earlier concerns shown about their massive civilian cost.
Dr. Nauman Ahmad, an IT and AI specialist in Lahore, tells Al Mayadeen Englishthat at the outset of “Israel’s” assault on Gaza, reservists were mass mobilized, and industrial professionals from the private sector began enlisting in the army espionage unit to bring in IA skills that had previously only been available to a tiny number of firms globally.
Telegram Will Start Moderating Private Chats, Altering Policy: Durov
“The reports emanating from media and private tech companies have already confirmed that Israel began making efforts to acquire espionage technology well before October 7; however, the Israeli army started hiring experts from tech giants such as Google, Meta, and Microsoft to establish an espionage dome to intercept Palestinian private conversations and develop language models much after October 2023,” he reveals, adding that they were now in the process of compiling and translating a huge amount of espionage data.
What is the new IT tool, and how does it analyze the data?
An investigation by the Associated Press uncovered that the Israeli military has ramped up its deployment of AI technology, incorporating tools from Microsoft and OpenAI, in response to the October 7, 2023, operation.
The report highlighted the risk of mass-scale human loss due to inaccurate data or defective algorithms. Although AI systems can accurately select targets, sometimes they make mistakes and blunders stemming from faulty feeding.
The findings prompt inquiries into Silicon Valley’s involvement in automated warfare and the consequences of technology’s role in facilitating unethical and unlawful military actions. The Israeli military’s use of AI has surged nearly 200-fold since October 2023, with the volume of data stored on Microsoft servers doubling from that period to July 2024.
The military’s utilization of Microsoft’s large computer servers increased by nearly two-thirds in just the first two months of the war. The consequences of this revelation could greatly influence the future of automated warfare, the report revealed.
Having gathered a substantial amount of spy data, Dr. Nauman stated that “Israel” is currently utilizing this collection of private conversations. He mentioned that they were creating an AI tool with the support of a cyber warfare team within the Israeli Military Intelligence Directorate.
He referred to this tool as a large language model (LLM) that analyzes information to generate, translate, and summarize conversational texts. The tool has been trained on 100 billion Arabic words, partially sourced from “Israel’s” extensive surveillance of Palestinians under military regulations, which he contends represents a significant infringement on Palestinian rights to privacy.
He stated that the Military Intelligence Directorate quickly recognized the necessity to create its program centered around the dialects that harbor animosity toward “Israel”. He explains that to achieve this, they trained the new LLM model on extensive intelligence about the private lives of Palestinians living under occupation.
“The current language models were limited to processing standard Arabic used in formal communications, while there was a need for a tool that could understand the spoken dialects of the Arab words,” he argued.
Why did Tel Aviv feel a need for a new AI gadget?
Since the beginning of the storm in Gaza, “Israel” put in place one of the most exhaustive reconnaissance mechanisms ever established in any conflict zone in recent history. Due to Hamas lacking an air defense system, “Israel’s” manned and unmanned reconnaissance operations can freely monitor all of Gaza, collecting valuable communication and electronic intelligence. Satellite imagery, ground forces transmitting operational information, and intelligence networking with its Western allies and the US provide them with stunning opportunities to amass a giant collection of sensitive information about the everyday lives of Gaza residents.
In a modern combat environment, the intelligence of the occupying forces is nearly flawless. However, “Israel” might struggle to transform its data into valuable insights because of the overwhelming amount of information it possesses.
This is where they felt a need to establish a new AI tool to minimize the reliance on human analysts and mitigate errors resulting from fatigue and overwork. Through pattern recognition and spoken dialect capability, the new AI machine can identify potentially valuable information from raw feeds and transform it into formats that are accessible to analysts. The processing capability of machine learning may allow analysts to draw more precise conclusions by integrating intelligence to create a comprehensive depiction of potential targets.
