
The Washington Post revealed that U.S. and Russian representatives revised the draft behind closed doors at the final stage of the United Nations lethal autonomous weapons negotiations in Geneva, Switzerland, and jointly deleted a number of key AI weapon safety regulations, including human review targets, predictable and reliable operations, and ethical considerations, triggering warnings from human rights organizations that it may lead to autonomous life and death decisions by machines and an increase in civilian casualties.
AI-generated summary
美俄在烏克蘭戰場上相互對抗,但在聯合國致命自主武器談判中罕見聯手修改草案,刪除多項AI武器安全規範,包括人類審查目標、可預測可靠運作及倫理考量條款。
United Nations Palace of Nations in Geneva, Switzerland. The Washington Post disclosed that U.S. and Russian representatives revised the draft behind closed doors in the final stage of negotiations on lethal autonomous weapons held here earlier this month, deleting a number of AI weapon safety specifications. (Reuters photo)
The United States and Russia, which are confronting each other on the battlefield in Ukraine, have rarely joined forces in an international negotiation at the United Nations on the regulation of lethal autonomous weapons. Hundreds of diplomats from around the world met in Geneva, Switzerland, earlier this month in an attempt to advance the first draft of an international agreement regulating the use of lethal autonomous weapons; however, informed sources and diplomatic documents revealed that representatives from the United States and Russia rarely joined forces in the final stages to revise the draft, jointly deleting a number of key safety provisions including "the target must be reviewed by humans before launching an attack."
The Washington Post exclusively reported on the 26th that the closed-door negotiations took place on the last day of negotiations. People familiar with the matter pointed out that during the intensive consultations that followed for about 15 hours, U.S. and Russian diplomats jointly made significant changes to the document, one by one deleting many of the safeguard clauses originally designed by each country to restrict lethal autonomous weapons.
Please read on...
According to people familiar with the matter, the US and Russian delegations deleted the language requiring AI weapon systems to operate in a "predictable" and "reliable" manner, and removed the clause that "ethical considerations" must be included when using AI weapons; more controversially, the two countries further deleted the requirement that AI-generated military targets must be reviewed by humans before launching an attack.
Camera closes, American and Russian lawyers make intensive revisions
People familiar with the matter pointed out that this modification process was carried out in a closed-door meeting with the United Nations' live cameras turned off and non-governmental organization observers invited away. The United States and Russia each sent about 10 legal experts, nearly twice the number of other delegations. The speed of revision was so fast that some delegations were overwhelmed. One participant even described the process as being "cut in pieces by a thousand knives."
The U.S. State Department, the Russian Foreign Ministry and the United Nations did not respond to requests for comment from The Washington Post.
The Washington Post pointed out that the weakening of this provision comes at a time when the U.S. Department of Defense is accelerating the integration of AI technology into battlefield strategies. The report mentioned that the application of AI in the military has gradually entered actual combat. Within 24 hours after the outbreak of the US-Iraq war, the US military used Anthropic's chatbot to identify nearly 1,000 targets; Israel's military operations in the Gaza Strip also rely heavily on AI systems.
Although OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei recently called for the establishment of a global AI regulatory framework at the United Nations, the Trump administration has strongly resisted international regulation. U.S. President Trump recently stated publicly at the United Nations General Assembly that the United States completely rejects any "globalist plan" that attempts to control AI.
Verity Coyle, deputy director of the Crisis, Conflict and Weapons Department of the human rights organization Human Rights Watch (HRW), warned that if the rules are weakened, it may allow machines to make life and death decisions without human control, causing more civilian casualties, reducing the level of accountability, and making war faster towards automation with higher risks.
Representatives from various countries are expected to discuss again in Geneva in November whether to obtain a formal authorization for relevant negotiations or to further promote a legally binding treaty.
AI outlook — possibilities, not facts
各國代表將於11月在日內瓦再度討論是否授權談判或推動具法律約束力的條約
Likely · Within months

Australia's home affairs authorities confirmed that Hongkongers pursuing study-to-migrate pathways cannot bring family members under new visa rules barring dependents on student or graduate visas, though families already in Australia will not be separated.

A military aircraft of the Democratic Republic of the Congo crashed due to mechanical failure on the way from Kikwit to Kinshasa on the 25th, killing 17 people, including the President of the Military High Court, Mutumbo, and the Auditor-General of the Armed Forces, Lieutenant General Likulia. The accident occurred near Kenji City, Kwango Province. The authorities originally counted 13 people dead. After the fire was brought under control on the 26th, the death toll rose to 17. Congolese President Tshisekedi ordered a thorough investigation.

Through the operation codenamed "Vevadi", the Ukrainian 3rd Army used large drones to airdrop unmanned ground vehicles (UGVs) loaded with explosives to the rear of the Russian-controlled area, about 19 kilometers deep. It has recaptured 5 villages and 3 settlements, resulting in the death of about 3,000 Russian soldiers and regaining 125 square kilometers of land.

A bus overturned on the Saveh-Hamadan Highway in Central Province of Iran on the 26th, killing 11 people and injuring 24 others. The Iranian traffic police pointed out that the cause of the accident was that the driver was drowsy due to fatigue driving. The number of traffic accident deaths in Iran throughout the year exceeded 17,000. The main causes of frequent traffic accidents include unsafe vehicles, traffic violations and insufficient ambulance.

German Foreign Minister Waldorf met with Russian Foreign Minister Lavrov on the sidelines of the United Nations General Assembly in New York. This was the first meeting between German and Russian foreign ministers since the outbreak of the Ukraine-Russia war. Waldorf demanded that Russia stop its dangerous escalation against Germany, Europe and NATO, and called for a deal on Black Sea grain exports. Lavrov said he had received no new information from the talks and suggested that Wadford call him directly or check his statement.

Chinese anthropologist Xiang Biao became popular after giving a speech in Singapore that focused on the anxiety of young people, and was hailed as "a star falling from the sky." He analyzed the sources of pressure at the turning point of Chinese society, emphasizing seeking meaning from others rather than institutional affirmation, and pointed out that the real estate and education systems amplified anxiety. The article mentioned that the youth unemployment rate once reached 21.3%, and words such as "lie flat" and "involution" are popular. Xiang Biao currently lives at the Max Planck Institute in Germany and advocates facing the pain points of the times to achieve analysis and management.