"US and Russia Led Efforts to Drop 'Human Review Safeguards' from UN AI Weapons Regulation Proposal"

The United States and Russia played leading roles in removing key clauses for the safe operation of artificial intelligence (AI) weapons from a United Nations proposal on lethal autonomous weapons (LAWs), the Washington Post reported on the 26th, local time.

According to the report, which cited anonymous sources, diplomats from the U.S. and Russia intensively revised the text during the final 15 hours of UN negotiations held in Geneva, Switzerland, earlier this month, deleting phrases requiring AI weapon systems to be operated in a "predictable and reliable" manner.

Provisions mandating ethical considerations when using AI weapons and regulations requiring human review before attacking military targets selected by AI were also removed, the WP reported.

Three anonymous sources familiar with the negotiations said that the U.S. and Russia deployed about 10 legal experts each during a closed-door meeting where the conference room cameras were turned off and civil society observers were asked to leave.

They mobilized nearly twice as many personnel as other delegations, rapidly rewriting the text, while smaller delegations struggled to respond, according to the report.

One source described the process as "a death of a thousand cuts," noting that numerous minor revisions gradually killed the consensus.

The U.S. State Department, the Russian Foreign Ministry, and the UN did not respond to inquiries from the WP.

The UN negotiations on lethal autonomous weapons are considered the international community's most advanced discussions aimed at regulating how AI-based weapons are used on the battlefield.

While they currently lack legal binding force, a consensus among member states could lead to a binding international treaty.

Countries are scheduled to gather again in Geneva in November to discuss whether to transition the negotiations into an official mandate procedure or a legally binding treaty.

Verity Coyle, senior advisor in the crisis, conflict, and arms division at Human Rights Watch (HRW), expressed concern that weakening the rules could allow machines to make life-and-death decisions without human control.

She pointed out that "civilian casualties could increase, accountability could be weakened, and we could rapidly move toward a more dangerous automated warfare."

Nicole Tan Hoeyen, secretary-general of the international NGO coalition Campaign to Stop Killer Robots, criticized that "the U.S. and Russia do not favor legally binding rules on autonomous weapons" and that they are "exercising excessive influence over the outcome of the negotiations."

The military application of AI is already becoming a reality.

The WP reported that the U.S. military utilized Anthropic's chatbot to identify about 1,000 attack targets during the first 24 hours of the war in Iran.

Israel is also known to rely heavily on AI in its military operations in the Gaza Strip.

The U.S. government is also revising its regulations on the use of autonomous weapons.

U.S. President Donald Trump issued a national security memorandum last June directing the Pentagon to update its autonomous weapons policy within 90 days, though the new guidelines have not yet been released.

The existing guidelines established under the Joe Biden administration in 2023 included provisions requiring human judgment to be applied to the review of autonomous and semi-autonomous weapon systems, as well as clauses mandating that AI weapon operators comply with the laws of war and relevant treaties.

The WP noted that these are similar to the clauses removed from the recent UN negotiation draft.

Coyle told the WP that "when seeking to preserve military and technological superiority, the U.S. pushes for voluntary guidelines instead of legally binding rules," adding that "they have used the same approach in cyber, space, and other arms control debates, and autonomous weapons are no exception."

Lethal autonomous weapons refer to weapon systems where AI analyzes sensors and video data to identify and track targets, and can perform some or all of the attack decisions or weapon operations.

A key issue for the international community is whether such systems can be prevented from making life-and-death decisions without meaningful human control.

(Photo: AP, Yonhap News)
※ Please note: This article was translated by AI and may contain errors.