First Instance of Fully Autonomous Drone Engaging and Killing Human Soldier

In a groundbreaking event, a fully autonomous drone has reportedly killed a soldier on the battlefield without any human oversight. This significant development, according to Ukrainian defense executives, signifies a pivotal moment in the ongoing conflict.

This unprecedented test involved deploying ten AI-operated Terminator drones along the front lines of the Ukraine war, resulting in the death of a Russian soldier.

“We tried it,” stated Alexander Kokhanovsky, the drone manufacturer responsible for the technology during an interview. New Scientist covered the remarks made at a press event held by the Embassy of Ukraine. “This was merely a test; we have never executed such an operation before.” [for broader perspective]

The test, conducted two years ago, involved a quadcopter drone programmed to advance towards the front lines, traversing 3 to 5 kilometers in roughly 10 minutes, before activating “terminator mode,” where an AI model identifies and engages targets.

“We simply launch the drone, and we know that everything in its vicinity will be neutralized. Everything the drone detects is targeted for destruction,” Kokhanovsky explained. “There’s no connection to the drone; you can’t view any video feed; whatever targets the drone identifies are eliminated.”

Due to the lack of visibility into the drones’ targeting systems, human-operated drones were later deployed to assess the area, which revealed casualties that included “several soldiers and a truck,” Kokhanovsky reported. Although there is no official record of the autonomous drones specifically targeting these victims, it was concluded they were responsible.

Kokhanovsky clarified that he did not directly participate in the tests, which were undertaken by unmentioned military units near Bakhmut and Chasiv Yar areas during a Ukrainian counteroffensive. The Ukrainian Ministry of Defense did not provide comments on the current legal status regarding testing and utilizing fully autonomous weapons.

While AI technology is increasingly adopted in military operations globally, enabling the efficient analysis of vast intelligence data and automating certain weapon functions, human involvement remains essential. Kokhanovsky’s admission represents the most concrete evidence to date indicating that combat-related deaths resulted solely from AI operations.

Officials from the defense industry stated at a press conference that despite current Ukrainian regulations forbidding the use of AI in critical target interception stages, numerous systems leverage AI throughout various phases of their operation. Kokhanovsky noted that the government acknowledges the rising potential of AI and is in discussions with defense manufacturers regarding possible regulatory changes.

A 2023 report indicated that some Ukrainian attack drones equipped with AI were autonomously identifying and striking targets, but these deployments primarily targeted vehicles rather than infantry. As of now, no human casualties have been confirmed.

Although an international ban on autonomous weapons capable of lethal actions without human oversight is not established, UN Secretary-General António Guterres expressed last year that: “There is no place in our world for autonomous lethal weapons systems.”

The United Nations has raised concerns regarding the potential of such weapons to eliminate human discretion from combat scenarios, potentially violating international humanitarian and human rights laws. There’s also a real possibility that autonomous systems might mistakenly engage friendly forces or civilians.

Numerous militaries have engineered technologies that automate various aspects of target engagement. The U.S. has developed software that compiles and evaluates extensive data to pinpoint battlefield targets for drone strikes, theoretically demanding human confirmation. There are also claims that the U.S. is advancing technologies for drones designed to autonomously identify and neutralize threats.

A 2021 United Nations report hinted that a Turkish-manufactured Kargu-2 quadcopter could have autonomously engaged human targets. This report noted an absence of specific details regarding the claims’ origins or confirmation of casualties, but did suggest that Libya’s Government of National Accord utilized drones against Haftar’s forces.

Major Danilo Polozhukno, an officer with Ukraine’s 21st Independent Unmanned Systems Regiment, who was unaware of or involved in the test, informed New Scientist that his battalion employs semi-autonomous systems, but human operators are consistently engaged.

“These drone systems and platforms ease the workload for operators by facilitating automatic target acquisition and tracking, along with autonomous guidance as they near the target. However, we do not incorporate fully autonomous systems that solely determine and engage targets without operator involvement,” Polozhukno remarked. “Ukraine is deeply committed to adhering to international humanitarian law and safeguarding the rights of all combatants, prioritizing decision-making that minimizes civilian casualties.”

Maria Rosalia Taddeo from the University of Oxford asserts that AI-assisted killings strip soldiers of their dignity and evade accountability for attackers, calling for a ban. “It’s more than just problematic; it’s alarming,” she commented. “Do we intend to foster a society where individuals can kill without human oversight, where the government permits such lethality?”

Anthony King, a researcher at the University of Exeter, stated that while fully autonomous attacks devoid of human participation are technologically feasible, their effectiveness may be overestimated.

“It’s entirely plausible that the government might endorse this for a military edge,” he noted. “Nevertheless, the actual usage of fully autonomous drones within the conflict remains minimal at best.”

“Thus, maintaining human oversight is not only morally imperative but militarily advantageous at this current juncture,” King asserted.

Kokhanovsky indicated that the Terminator project has stalled since the test due to Ukrainian regulations. As the CEO of Aerocenter, a Ukrainian drone manufacturing firm, he clarified that he was not involved in the tests as the company had not been established at that time. Aerocenter focuses on developing autonomous interceptor drones intended to target and neutralize incoming Russian Shahed suicide drones before they can reach populated areas and critical infrastructure.

The ALITA system developed by the company boasts 16 launch pads equipped with 64 drones. Set for completion by October, it aims to autonomously monitor and engage incoming threats at speeds of 450 kilometers per hour, capable of intercepting small drones and helicopters alike.

However, existing Ukrainian regulations restrict fully autonomous operations, necessitating human identification of targets during final interception phases. This protocol enables a minimal two-person crew to operate 64 drones seamlessly, significantly reducing manpower.

“All stages of this operation can be executed manually or automatically, but the final engagement must be manually conducted,” Kokhanovsky emphasized, advocating for a change in regulations. “I’d advocate for that shift,” he concluded.