Guterres Calls for Ban on Autonomous Weapons at UN Summit in Geneva

António Guterres put it bluntly: machines with the power and discretion to take human lives without human control should be prohibited under international law. The UN Secretary-General made the demand on July 8 at the first Global Dialogue on AI Governance in Geneva, where representatives from 193 countries gathered.

Guterres's proposal follows a two-tier structure. The first tier is an outright ban on weapons systems that select and attack targets on their own without meaningful human control. The second tier is strict regulation of other autonomous weapons that don't make kill decisions entirely independently but use AI for other combat functions.

Three Risks Identified by the Science Panel

The Independent International Scientific Panel on AI, which prepared a report for the summit, identified three main risks. The first is the pace of algorithmic development outstripping governments' ability to control it. The second is the concentration of digital power in the hands of a small number of corporations. The third is generative technology's ability to convincingly impersonate reality, undermining public trust and democratic processes.

Guterres described the current stage of AI development as a giant experiment being run on societies without citizens' informed consent. He added that the current generation may be the last with a chance to set the terms for how humans and machines coexist.

Machines that have the power and discretion to take human lives without human control should be prohibited by international law - António Guterres, UN Secretary-General

The Accountability Gap

The key legal problem experts point to is the so-called accountability gap. Current international law has no mechanism for assigning blame when an autonomous system kills an unintended target. It remains unclear whether responsibility falls on the commanding officer, the software maker, or the state that deployed the system.

Human Rights Watch, which has campaigned for years to ban so-called killer robots, backed Guterres's position. The Stop Killer Robots campaign has been pushing for a preemptive ban since 2013, but only now has the issue reached the highest diplomatic level at the UN with a concrete deadline attached.

Four Priorities and US Opposition

Beyond the weapons ban, the UN outlined four priorities for AI governance: common safety standards, inviolable human rights protections, support for developing countries in building their own technological capacity, and full transparency from companies developing AI systems. On top of that came a call for a global commitment to child safety, including rigorous testing of systems before deployment and a zero-tolerance policy on content exploiting minors.

The United States remains the main obstacle to a binding treaty. Washington consistently favors voluntary codes of conduct over legally binding bans, arguing that autonomous systems can be more precise than frightened soldiers and can reduce civilian casualties. Some military ethicists share this view, arguing that combat robots could lift part of the moral burden of waging war off soldiers' shoulders.

What It Means for Poland

As a NATO member and participant in the AI Act discussions, Poland has a direct stake in the outcome of these negotiations. The development of autonomous systems on NATO's eastern flank, where technologies such as Palantir's platforms already support tracking troop movements, means the question of how far autonomy in weapons should go is no longer theoretical. Any eventual treaty would also affect Poland's domestic defense industry, which is developing unmanned systems.

Guterres announced that further rounds of negotiations will take place later this year, with the UN aiming to have the treaty adopted before the end of 2026. Given resistance from the United States and Russia to binding regulation, experts view that deadline as highly ambitious, but they see the very fact that the issue reached a summit with 193 countries as a breakthrough in a discussion that had previously played out mostly among narrow expert circles.

Sources: Semafor (semafor.com), Rzeczpospolita Cyfrowa (cyfrowa.rp.pl), Yahoo News (yahoo.com)