This post is also available in: heעברית (Hebrew)

There is a growing development and deployment of AI-controlled drones that can autonomously select and attack human targets. The US is reportedly developing a network of such drones that could be rapidly deployed near enemy regions.

These drones are also known as lethal autonomous weapons or “killer robots” and are being pursued by several countries despite the ethical and legal concerns that such weapons could pose a grave threat to humanity and international security. The Times reported that many governments are urging the UN to adopt a binding resolution that would ban or limit the use of AI killer drones. Nevertheless, several countries, including the US, Russia, Australia, and Israel, insist that they prefer a non-binding resolution.

Air Force secretary Frank Kendall said that AI drones would have to be able to make lethal decisions under human supervision, adding that he did not think that the adversaries of the US would refrain from making individual decisions and that it would give them a huge advantage if the US imposed that limitation on itself.

Ukraine had reportedly already used AI-controlled drones in its conflict with Russia, but it is unclear whether these attacks had caused any human casualties.

It is important to note that the concept of autonomous weapons is not new. Beyond land mines that have been used since the 1800s, a more recent example is homing munitions, which generally cannot be recalled after they are fired and have a certain degree of autonomy in refining their path. Harpoon anti-ship missiles operate similarly, with limited autonomy. Another example is loitering munition, which can stay in the air for a long time, searching for a specific target type and then attacking it.

According to Interesting Engineering, the Pentagon is developing a network of hundreds or even thousands of AI-enhanced, autonomous drones that could be rapidly deployed near China in the event of conflict. These drones would carry surveillance equipment or weapons and would be used to take out or weaken China’s extensive network of anti-ship and anti-aircraft missile systems along its coasts and artificial islands in the South China Sea.

The Pentagon reportedly aims to deploy thousands of inexpensive, autonomous, and even lethal drones in the next year or two that can continue operating even when GPS signals and communications are jammed.

Nevertheless, military contractors argue that fully autonomous AI-controlled lethal attacks are still years away since the most advanced algorithms are not yet reliable enough and cannot be trusted to make life-or-death decisions.