Users Prompted Grok to Simulate Israeli Targeting System
Social media users are leveraging the chatbot to generate risk scores for accounts based on public posts.
Updated on Sept. 21, 2026 in Artificial Intelligence
Live Poll
Do you trust AI chatbots to accurately assess the intent or risk level of social media users?
Users on X have been utilizing the Grok chatbot to simulate the function of Lavender, an AI-assisted system reported in 2024 to have been used by Israeli intelligence. These prompts generate numerical concern scores ranging from 1 to 100 based on an analysis of public post history.
Why it matters
The practice highlights how public-facing large language models can be steered to mimic sensitive military workflows, despite having no access to classified intelligence. The tool is being used by participants to evaluate public social media activity through a lens of simulated risk assessment.
Grok assigns concern scores on a scale of 1 to 100 by parsing account history for criticism of Israel. The system operates entirely on public posts on the X platform rather than accessing classified intelligence or actual Israeli military software.
The players
Grok
An AI chatbot integrated into the X social media platform that processes public posts to generate conversational responses.
Israeli military
The national armed forces that have denied using AI systems to independently determine military targets.
The details
Users prompt the chatbot to analyze accounts using the nomenclature of Lavender, an AI-assisted system first reported in 2024. The model performs sentiment and pattern analysis on publicly available post data to produce risk metrics. While reports have suggested that Lavender was used to flag tens of thousands of people during the war, the Israeli military denies employing any AI system that independently determines lethal targets.
Timeline
2024: The Lavender system was first widely reported.
September 21, 2026: This article was published.
The Tech Race
This activity follows a pattern set by the 2024 reports regarding the use of the Lavender AI system in Gaza. It reflects how users are attempting to project high-stakes military software capabilities onto widely accessible commercial LLMs.
The simulation uses public-facing data on X to generate results for any account visible to the model. Users should remain aware that these scores are generated by a standard LLM and lack any connection to classified intelligence or actual military operational frameworks.
The takeaway
The use of Grok to mimic sensitive military software demonstrates a growing public interest in testing the boundaries of LLM applications. Future observers should monitor if platforms implement specific content filters to prevent the use of their tools for simulating military-grade target identification.
Further reading
For more background on the deployment of large language models, visit Artificial Intelligence.
Source note: This article includes information reported by ProPakistani.
Live Poll
Do you trust AI chatbots to accurately assess the intent or risk level of social media users?
