Israeli military developed ChatGPT-like AI tool trained on Palestinian surveillance data
Israel's military surveillance agency Unit 8200 developed a large language model similar to ChatGPT trained on intercepted Palestinian communications to analyze surveillance data. Experts have expressed concern about potential biases and misuse, including tracking human rights activists and expanding surveillance across Palestinian territories.
What happened
Israel's military surveillance agency Unit 8200 built a large language model (LLM) similar to ChatGPT trained on vast volumes of intercepted Palestinian telephone conversations and text messages to create a chatbot-like tool for analyzing surveillance data.
The unit began developing the system after the start of the war in Gaza in October 2023, and the model was still being trained in the second half of 2024.
A former military intelligence technologist, Chaked Roger Joseph Sayedoff, publicly described the project at a military AI conference in Tel Aviv, stating the team tried to collect all Arabic language data the state of Israel possessed and required "psychotic amounts" of data.
Reported impact
Sources described how the technology could be used beyond counterterrorism, including to track human rights activists, monitor Palestinian construction in Area C of the West Bank, and provide surveillance tools to know what every person in the West Bank is doing.
Experts noted that such AI systems can exacerbate biases and are prone to making mistakes.
Index analysis
The AI model was trained using large-scale retention of intercepted communications obtained through Unit 8200's extensive surveillance of occupied Palestinian territories.
It is unclear whether the model has yet been deployed operationally.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.