The United States and Russia have forced the removal of key safeguards from a proposed United Nations treaty on lethal autonomous weapons, including a requirement that humans review military targets selected by artificial intelligence before an attack, the Washington Post reported on September 27.T
The changes were made during closed-door negotiations in Geneva earlier this month under the UN Convention on Certain Conventional Weapons (CCW) — the furthest-advanced international effort to set rules for weapons that can identify, select and engage targets with limited or no human intervention.

Fifteen hours that rewrote the draft

On the final day of talks, American and Russian diplomats spent roughly 15 hours revising the document, working from a conference room with the UN cameras switched off and civil society observers asked to leave, the report said, citing three people familiar with the negotiations and documents reviewed by the newspaper.
The two delegations removed language requiring autonomous weapons systems to operate in a "predictable" and "reliable" manner, deleted a clause calling for ethical considerations in how such systems are used, and struck the provision requiring human review of AI-generated military targets before a strike.
References to the design and development of autonomous weapons were also cut, along with terms covering explainability and traceability — safeguards meant to show how an AI system reaches a targeting decision and to assign accountability when errors occur.

A diluted but advancing framework

Human Rights Watch said Russia and the United States weakened provisions that were widely supported by other governments, and noted the legal framework was narrowed from international law broadly to international humanitarian law specifically.
The framework that emerged still addresses human judgment, control, predictability, reliability, traceability and explainability, but in softened form.
Both Washington and Moscow have consistently favoured national guidelines for autonomous weapons over a binding international treaty, and 76 nations already back converting the nonbinding text into a binding one, according to the rights group.
The US State Department, Russia's foreign ministry and the United Nations did not respond to requests for comment from the newspaper.

Why it matters beyond Geneva

The outcome lands amid growing deployment of AI-assisted systems by major militaries, and it echoes an unresolved debate inside the Pentagon:
President Donald Trump signed a memorandum in June directing it to update Directive 3000.09, the policy requiring human judgment before autonomous weapons fire, within 90 days. As of Saturday, that revision was still overdue.
For the AI industry, the diluted text leaves developers of dual-use systems facing fewer international constraints — and a global patchwork of national rules instead of one binding standard.
By Hamza Faisal