Israel's military intelligence division has developed a capable ChatGPT-like artificial intelligence system, based on millions of intercepted Palestinian communications. The system was developed by Unit 8200 and it claims that the system will enhance intelligence but raises issues of privacy, bias, and abuse.
Israel's military intelligence division has created an artificial intelligence program like ChatGPT based on a huge cache of intercepted Palestinian communications, an investigation by The Guardian, +972 Magazine, and Local Call found. The AI model was developed by Unit 8200 and trained on vast amounts of phone calls and text messages harvested through widespread monitoring of the occupied territories.
A new frontier in military surveillance
The AI system, which is still in the works, is intended to act as a sophisticated chatbot, responding to queries regarding people being monitored and interpreting huge amounts of data gathered. Israel's top eavesdropping unit, Unit 8200, stepped up the project after the Gaza war broke out in October 2023. People close to the project indicate that the model was being trained through to late 2024, although it is not certain whether it has been deployed fully.
Three former intelligence officials confirmed the existence of the AI model, which they described as an improvement on earlier machine-learning systems employed by Unit 8200. One of the sources mentioned that such AI-facilitated surveillance enables intelligence officers to monitor human rights activists, observe Palestinian building in the West Bank, and monitor everyday life in the occupied territories.
Bias and accuracy concerns
The creation of the AI tool is a serious ethical and legal issue, especially in its ability to perpetuate biases and commit mistakes in intelligence analysis. Human rights organizations have cautioned that the application of large language models (LLMs) in warfare can result in erroneous conclusions that affect the lives of Palestinians living under occupation.
Zach Campbell, a senior surveillance researcher with Human Rights Watch, was alarmed by the capabilities of the system. "It's a guessing machine, and in the end these guesses can end up being used to implicate people," he said.
Israel's military has not provided information on how it verifies accuracy in AI-generated intelligence. An Israel Defense Forces (IDF) spokesman said only that the military "employs various intelligence means to detect and disrupt terrorist activity by hostile organizations in the Middle East."
The use of US tech expertise
Unit 8200's initiative was supported by reservists who had previously served in the private sector, including talent from large US tech companies like Google, Meta, and Microsoft. They assisted in sharpening the AI model to interpret spoken Arabic more effectively, a demanding task considering the lack of large-scale training data for local dialects.
Sayedoff, the ex-intelligence official, explained the challenge of constructing such a model. "There are no call transcripts or WhatsApp discussions on the internet in quantities sufficient to train such a model," he stated. In response, Unit 8200 collated decades' worth of overheard discussions into a single database, amounting to an estimated 100 billion words.
Several retired Western intelligence officers have claimed that Israel's deployment of AI for surveillance is more than is tolerable within US or UK regulation systems. While intelligence organizations in both nations are subject to strictures against information gathering and processing, Israel's Unit 8200 works under less such restriction.
AI-driven surveillance and its risks
The increasing use of AI in military operations has already begun to shape Israel's war strategies. The Gospel and Lavender systems have been instrumental in identifying targets for airstrikes in Gaza. These AI systems, according to sources, assisted in the creation of kill lists utilized by the IDF during bombings.
Although it has the capability to process huge amounts of data, AI technology is not immune to error. AI practitioners caution that machine-learning algorithms have the capacity to produce questionable conclusions, with possibly catastrophic effects on military decision-making.
Implications for human rights
Digital rights groups have criticised the application of AI in military espionage, contending that Palestinians essentially have become a test population for an experimental security system. 7amleh director Nadim Nashif opined that the technology is deployed "to control a people, to dominate their lives.
Unit 8200's AI-powered surveillance has already resulted in more arrests in the West Bank, according to sources. One former intelligence officer explained how the AI system assists in monitoring people based on speech patterns that are considered "troublesome."
"AI enables us to act on information regarding many more individuals, providing the IDF with greater control over the population," a former intelligence official explained.
A new era of AI-driven intelligence
Unit 8200's big language model is a huge leap toward AI-based intelligence collection. Comparable projects are being conducted in the US and UK, where the two countries' intelligence agencies are working on creating their own AI applications to sort through enormous collections of data. Experts, though, say Israel's military seems to be extending the limits of AI monitoring even beyond its closest allies.
As AI becomes more and more central to military operations, worries about its dependability, moral implications, and human rights implications will only increase. Israel's AI-powered espionage experiment may provide a model for future global intelligence operations—one that critics caution will be towards more surveillance and less accountability.
Discover the latest Business News, Sensex, and Nifty updates. Obtain Personal Finance insights, tax queries, and expert opinions on Moneycontrol or download the Moneycontrol App to stay updated!
