Automated warfare raises alarm over AI-driven targeting in Gaza

Since May 2021, when the Israel Defense Forces launched what its officers described as the “first AI war” during the 11-day conflict in Gaza, Israel has continued to expand its use of artificial intelligence technology to monitor and target Palestinian civilians.

The escalation of military tactics to include automated systems that operate with minimal human intervention has raised serious concerns about reliance on unauthorized data extraction and computerized decision-making in paramilitary operations, demonstrating the growing convergence of mechanized warfare and information capital in the digital age.

Despite Israeli Defense Force officials, including former Chief of Staff Aviv Kochavi, claiming that these applications “generate vast amounts of data more efficiently than humans,” a recent review of Israel's use of the software found that the algorithms were unreliable and unsuitable for decisions as critical as determining the life or death of an individual.

A 2024 assessment by Human Rights Watch of Israel's use of four AI-powered systems concluded that Israel relied on “incomplete data and imprecise approximations to inform military operations,” highlighting their propensity for errors and contradictions despite official assurances to the contrary.

The IDF's use of AI systems to determine military targets also raises questions about the scope of human surveillance. One system, known as “Lavender,” uses machine learning to assign Gazans a numerical score indicating their likelihood of being part of an armed group, Human Rights Watch said.

The IDF said the tool is used in parallel to manual procedures performed by military officers and is not meant to replace the role of intelligence analysts.

In a 2024 press release clarifying the use of AI tools, the military said: [Lavender] “We cannot rely on the database alone to identify him as a military operative who could be attacked,” it said, adding that artificial intelligence tools “are not the sole basis for determining who to attack.”

However, reports from IDF officials suggest that human involvement in reviewing AI-generated materials is limited beyond a simple evaluation of results.

One police officer who used Lavender told the Guardian: “At this stage you would spend 20 seconds on each target and do dozens of them every day. Apart from the seal of approval, there was zero added human value.”

Despite IDF claims that the system operates within a broader target verification process, this statement suggests a narrow form of surveillance.

The 2024 Human Rights Watch report further warned that military reliance on such systems could violate international humanitarian law, particularly the laws of war, which distinguish between military objectives and civilians and require all possible precautions to be taken to minimize civilian casualties.

The report characterizes Israel's use of AI technology as an extension of extractive surveillance practices that collect personal data from Palestinian civilians in violation of international human rights law and feed that information into machine learning systems aimed at distinguishing civilians from military purposes.

According to the report, these tools appear to rely more on speculative assumptions drawn from illegally obtained data than on rigorous technical procedures to determine the fate of Gazans. As long as AI programs continue to be used to identify targets without meaningful oversight, the military risks violating the laws of war, which require the military to always carefully distinguish between civilians and combatants to limit civilian casualties.

The UN Special Committee suggested that the use of AI with limited human oversight by the Israel Defense Forces may reflect a disregard for these obligations, stating in a 2024 press release that “Israel's military's use of AI-assisted targeting in conjunction with minimal human oversight and heavy bombs highlights Israel's disregard for its obligations to distinguish between civilians and combatants and to take appropriate security measures to prevent civilian deaths.”

The IDF's adoption of these technologies reflects a broader shift to automated or unmanned systems for generating targets in paramilitary operations.

In The Theory of Drones, French theorist Grégoire Chamayou explains how drone warfare collapses surveillance and combat into a single mechanism, writing: “Here, vision is vision. It does not represent objects, but acts on them and serves to determine targets. The function of the eye is the function of weapons.”

Chamayu argues that the images we see on screen are not objective reflections of reality, but rather projections shaped by subjective assumptions embedded in technology.

The targeting process relies on identifying algorithmically defined “non-normative” behavior rather than recognizing individual identity, Chamayou writes.

“This information model paradoxically claims to be able to ‘identify’ individuals while remaining anonymous, as it presupposes the analysis of behavioral patterns rather than the recognition of nominal identities,” he says. Such identification, he argues, is not personal but general.

This inherent bias, masked by claims of technological neutrality, is evident in the IDF's AI-powered system, which processes data from Palestinians in Gaza and “draws inferences from it.” [such] Retrieve data and recognize patterns without explicit instructions. ”

By reducing people to statistical profiles and generalized characteristics while maintaining objectivity, these systems undermine the reciprocity of combat and eliminate the possibility of self-defense. The entire life and personal history is reduced to technical variables, and war becomes a logistical operation that does not require direct knowledge of human goals.

As the death toll in Gaza continues to rise, with more than 69,000 Palestinians reported killed and more than 170,000 injured since October 7, 2023, the IDF's use of AI systems to rapidly generate targets has raised concerns that these tools are accelerating and escalating the loss of life, according to a UN humanitarian update citing information from the Gaza Ministry of Health.

Opportunities for mutual self-defense have been effectively erased, and Palestinian lives are increasingly subject to automated decision-making, with their survival dependent on the output of algorithms.

Follow Vanguard on social media including X, Instagram, and Facebook. Subscribe to the Vanguard Newsletter. Click here to make a tax-deductible donation. davisvanguard.org/donate or give directly act blue. Your support ensures Vanguard's important work continues.