2026年6月23日 晚上9:38 UTC / 路透社
2023年9月20日，印度孟买一场会议期间，人们走过Meta平台公司的标识。路透社/弗朗西斯·马斯卡雷尼亚斯 获取授权许可 打开新标签页
6月23日（路透社）——据《纽约时报》周二援引四名熟悉该机密请求的人士消息，特朗普政府正施压Meta自愿提交其AI模型供审查，此举将允许美国政府评估这些模型的能力与漏洞。
报道称，这项通过电子邮件向这家社交媒体巨头提出的请求，正值本届政府加大对AI行业监管力度之际。
订阅路透社美国政治新闻简报，获取每周美国政治新闻及分析，了解政治如何影响全球。点击此处注册。
广告 · 继续向下滚动
举报广告！
据报道，这家今年4月推出Muse Spark AI模型的脸书母公司，是美国唯一尚未达成自愿与联邦政府共享模型以供审查协议的大型AI技术开发商。
“我们与本届政府有着相同目标，即推动美国在稳健且安全的前沿AI领域占据领先地位。目前我们正在敲定细节，希望能尽快签署协议，”Meta在给路透社的电子邮件回复中表示。
美国商务部未立即回应路透社的置评请求。
本月早些时候，美国政府以国家安全担忧为由，要求Anthropic暂停外国用户访问其最先进的AI模型。
广告 · 继续向下滚动
OpenAI和Anthropic此前已在与美国政府合作测试未发布的AI模型，而谷歌DeepMind、微软和xAI则在5月同意向政府提供新模型的早期访问权，用于国家安全评估。
华盛顿方面对强大AI系统构成的国家安全风险的担忧与日俱增。通过提前获取前沿模型，美国官员旨在抢先识别从网络攻击到军事滥用等各类威胁，防止这些工具被广泛部署。
6月2日，美国总统唐纳德·特朗普签署一项行政命令，建立自愿框架，要求AI开发商在向可信合作伙伴发布“受覆盖前沿模型”前，先向美国政府提交审查，时长最多30天。
班加罗尔的贾斯普雷特·辛格 报道；萨哈尔·穆罕默德和维杰·基肖尔 编辑
我们的准则：汤森路透信任原则 打开新标签页
US presses Meta to agree to AI reviews as security concerns rise, NYT reports
June 23, 2026 9:38 PM UTC / Reuters
People walk behind a logo of Meta Platforms company, during a conference in Mumbai, India, September 20, 2023. REUTERS/Francis Mascarenhas Purchase Licensing Rights, opens new tab
June 23 (Reuters) – The Trump administration is pressing Meta to submit its AI models for voluntary review, which would allow the government to evaluate their abilities and vulnerabilities, the New York Times reported on Tuesday, citing four people familiar with the confidential request.
The request was made in emails with the social media giant, the report said, as the administration steps up oversight of the AI industry.
Get weekly news and analysis on U.S. politics and how it matters to the world with the Reuters Politics U.S. newsletter. Sign up here.
Advertisement · Scroll to continue
Report Ad!
The Facebook parent, which launched the Muse Spark AI model in April, is the only major U.S. developer of AI technology that has not reached an agreement to voluntarily share its models with the federal government for review, according to the report.
“We share the administration’s goal of advancing U.S. leadership on robust and secure frontier AI. While we are working through the details, we hope to sign the agreement soon,” Meta told Reuters in an emailed response.
The U.S. Commerce Department did not immediately respond to a Reuters request for comment.
Earlier this month, the U.S. government ordered Anthropic to suspend access to its most advanced AI models for foreign nationals, citing national security concerns.
Advertisement · Scroll to continue
OpenAI and Anthropic had already been working with the U.S. government to test unreleased AI models, while Google DeepMind, Microsoft and xAI agreed in May to provide the government early access to new models for national-security evaluations.
Concern is growing in Washington over the national security risks posed by powerful AI systems. By securing early access to frontier models, U.S. officials are aiming to identify threats ranging from cyberattacks to military misuse before the tools are widely deployed.
On June 2, President Donald Trump signed an executive order establishing a voluntary framework for AI developers to offer “covered frontier models” to the U.S. government for up to 30 days before releasing them to trusted partners.
Reporting by Jaspreet Singh in Bengaluru; Editing by Sahal Muhammed and Vijay Kishore
Our Standards: The Thomson Reuters Trust Principles., opens new tab
发表回复