In the grinding war of attrition between Russia and Ukraine, unmanned aerial vehicles of many kinds now cause 70 to 80 percent of all deaths and wounds on each side.

The demands of prolonged fighting and wider UAV use have formed a repeating loop. With drones everywhere creating larger kill zones, both forces now rely on small mobile units for advances or recoveries, which makes major gains almost impossible. This situation has sped up UAV improvements, focused mainly on electronic warfare and jamming of control signals.

Although neither side created these tools, both have built or refined two main answers to jamming. One uses a fibre-optic cable to link the operator and UAV, shielding it from radio interference. The other employs AI to pick and strike targets without any human input, removing any link that could be jammed.

The cable method has already produced lasting environmental harm. The AI method, known as lethal autonomous weapons systems or LAWS, carries risks that reach well beyond nature. Both approaches harm humanity's future, yet global rules trail behind, especially on LAWS.

No shared international definition of LAWS exists, but major groups describe them as systems that, once switched on, choose and use force based on sensor data instead of human commands.

Human Rights Watch first drew this line in 2012 and argued for an early ban on killer robots, noting that LAWS would weaken responsibility for deadly choices since no coder, maker, or officer could be held accountable for illegal deaths.

Because LAWS represent a major technical shift, some call them the Oppenheimer moment of this century, yet they create a larger problem than classic mass-destruction arms.

Chemical and biological arms can work in narrow settings but prove unreliable and risky to the user. Nuclear arms bring huge destruction, yet their use today equals global self-destruction due to wide spread.

The real threat from LAWS is not raw power but the absence of traits that once kept mass weapons in check, making them tougher to limit.

They cost little and can be made without state help. Their core tech serves both military and civilian ends, dodging export rules. Most critically, their target rules sit inside software that stays hidden, deniable, and impossible to tally like warheads.

Efforts to regulate LAWS globally also face pushback from heavily armed nations.

The UN General Assembly passed three resolutions on LAWS from 2023 to 2025, and the short lists of opposing states are revealing.

In 2023 opponents included Belarus, India, Mali, and Russia; in 2024 they were Belarus, North Korea, and Russia; in 2025 those three added Burundi, Israel, and the US.

China abstained each time. It often speaks of rules for LAWS but aligns with Russia and the US by rejecting firm limits on the very systems it is building fast.

Dissonance, defiance, and hypocrisy

Even while backing the UNGA resolutions, the EU's stance on regulating killer drones stays weak and uneven.

Austria became the first Western democracy in 2018 to join calls at the UN Convention on Certain Conventional Weapons for a LAWS ban and later introduced the first UNGA resolution in 2023.

France, however, has worked since 2022 to soften EU AI rules and exclude military, defence, or security systems, which would cover LAWS.

The EU as a body still claims the CCW is the right place for LAWS talks, aware that its consensus rule and Russia's presence have stalled all progress.

With Brussels' mixed signals, Beijing's double talk, and open resistance from Moscow and Washington to the global majority, binding rules on LAWS seem unlikely. The Russia-Ukraine war, where drones with autonomous features already kill, offers ideal training data for AI targeting humans.

Chemical arms were banned after the First World War horrors, but optimism is harder with killer robots: if they fail, supporters will simply call it a coding issue for the next version.

Russia holds exactly this view, claiming LAWS can outperform human operators and reduce mistakes.

Yet mistakes are not the core issue. The real problem is handing the choice to kill over to a machine in the first place.