When diplomats gathered in Geneva to finalize the first international rules on lethal autonomous weapons, the United States and Russia spent the last day taking them apart. As The Washington Post’s Pranshu Verma reports, U.S. and Russian negotiators worked for roughly 15 hours to strip key safeguards from the draft, according to three people familiar with the closed-door talks.
Out went language requiring AI weapons to operate in a “predictable” and “reliable” manner. Out went a clause mandating ethical considerations. And out went a provision requiring humans to review AI-generated military targets before a strike.
With U.N. cameras off and civil society observers removed, each country fielded about 10 lawyers — nearly double other delegations — and moved so fast that smaller teams were left dizzied. “It was death by a thousand paper cuts,” said one person familiar with the negotiations.
The stakes aren’t hypothetical. The U.S. military used Anthropic’s chatbot to identify nearly 1,000 targets in the first 24 hours of the war with Iran, and Israel has leaned heavily on AI in Gaza.
“It could mean machines can make life‑and‑death decisions without human control,” said Verity Coyle, deputy director for arms in the crisis, conflict and arms division at Human Rights Watch. “The practical impact is stark: more civilian harm, less accountability and a faster slide into riskier, automated warfare.”
Related on the SLP
The talks are nonbinding, but nations reconvene in November to weigh a binding treaty. Nicole van Rooijen, executive director of the coalition Stop Killer Robots, warned that Washington and Moscow “have the power to block any significant progress that is made by the vast majority of states who do want to move forward on a binding treaty.”
Meanwhile, the Pentagon is rewriting its own 2023 autonomous weapons directive — which includes the very human-judgment requirements just scrubbed from the U.N. text. President Trump ordered the update in June; as of late September, no new version has been released.
