OpenAI工具卷入黑客事件后,特朗普考虑加强AI管控
Trump considering AI controls after OpenAI hacking incidents
BBC News · Technology & AI · C1 Advanced · 2 min read · 2026-07-30
摘要
在OpenAI承认其工具涉及黑客事件后,特朗普表示美国正考虑加强AI管控,但同时希望保持对华竞争优势;美国政府近期也加大了对中国科技企业的施压力度。
President Trump said the US is considering tighter AI controls after OpenAI took responsibility for hacking incidents, while seeking to preserve America's competitive lead over China and increasing pressure on Chinese technology.
长难句拆解
Asked about OpenAI's tools being responsible for improperly breaching the private technology of other companies, Trump said: "We're looking at AI, we're looking at controls, we're also making sure that we lead."当被问及OpenAI的工具不当侵入其他公司的私有技术并应对此负责一事时,特朗普说:“我们正在研究人工智能,也在研究管控措施,同时还要确保我们保持领先。”
主干:[When he was] asked about A, Trump said: “We’re looking at AI; we’re looking at controls; we’re making sure that we lead.”
先抓住两层关系:开头被问话的人是特朗普;said 后面的引语包含三个并列陈述,最后一个陈述内部又带有 that 从句。
- 句首 Asked 是压缩后的被动背景:Asked 不是 Trump 主动“询问别人”,而是“Trump 被问”。英语常把 When he was asked 压缩成 Asked,并把它放在主句前交代说话背景。
- about 后面装入了一个完整的事情:about 的内容不只是 OpenAI's tools,而是“这些工具要为某种行为负责”这一整件事。tools 是 being responsible 的主体,也被理解为 breaching 的主体。
- responsible for 后用 -ing 表示所负责任的行为:responsible for 后面的 breaching 指具体要承担责任的行为;improperly 修饰 breaching,表示这种侵入是不当的。
- 引语里有三个并列陈述:三个部分都以 we 为主语,分别表达“研究 AI”“研究管控措施”和“确保领先”。原句用逗号连续列出,also 标明第三项是在前两项之外增加的内容。这里的 we 结合上下文主要指特朗普及其政府,而不是采访者。
- 进行时表示当前正在推进的政策工作:三个进行时都以特朗普说话的时点为中心,表示相关审视和保障工作正在进行,可能此前已经开始。look at 在这里不是“看一眼”,而是“研究、评估”;making sure 表示采取行动确保某个结果。
- that we lead 是 make sure 的具体内容:that 从句不能与前三个陈述并列;它只依赖 making sure,说明要确保的结果。lead 在这里不带宾语,意为在相关竞争中处于领先位置。
The US government intervened when Anthropic, OpenAI's main American rival, decided to release to the public a model that it had previously said was too much of a risk to be made widely available.当OpenAI在美国的主要竞争对手Anthropic决定向公众发布一款模型时,美国政府进行了干预;Anthropic此前曾表示,该模型风险过高,不宜广泛开放。
主干:The US government intervened when Anthropic decided A.
先抓住两件事:美国政府“干预”;when 从句说明触发干预的事情是 Anthropic 决定公开一个模型。模型后面还有一层说明:Anthropic 以前认为它风险太大。
- 主句与 when 从句:干预由什么触发:主句回答“谁做了什么”:美国政府进行了干预。when 后面回答“在什么情况下”:Anthropic 作出发布决定时。这里的 when 不只是钟表时间,也把该决定呈现为政府介入的触发背景。
- 逗号中的同位说明:OpenAI's main American rival 与 Anthropic 指同一家公司,只是补充其身份。删除这段后,句子的主要结构仍然成立。
- decide to do:决定采取某个行动:to release 是 decided 的内容。to the public 表示发布对象是公众;原句把它放在 release 后、a model 前。
- 远距离关系:that 最终与 was 相连:that 指 a model。虽然中间插入了“it had previously said”,但核心关系仍是“the model was too much of a risk”。it 指 Anthropic。因此意思是:Anthropic 以前说,这个模型风险太大。
- had previously said:更早发生的评价:had said 是过去完成时,表示“说风险太大”发生在另一个过去事件“决定发布”之前;previously 又明确强化了这个先后关系。
- too much of a risk 与被动开放:too much of a risk 表示风险大到不可接受。to be made widely available 使用被动形式,关注“模型被广泛开放”这一结果,而不强调由谁开放。整体意思是:风险太大,因此不适合让它得到广泛使用。
Most AI models being developed in China are open source, meaning they are built and then made freely available online for anyone with the proper computer equipment to download and use.目前在中国开发的大多数AI模型都是开源的,也就是说,这些模型开发完成后会在网上免费开放,任何拥有适当计算机设备的人都可以下载和使用。
主干:Most AI models [that are] being developed in China are open source, meaning they are A and then B.
先抓住主干:“大多数AI模型是开源的”。meaning之后解释“开源”具体意味着什么,其中built和made是两个先后发生的并列动作。
- being developed是缩略的进行时被动修饰语:它修饰AI models,完整理解是“正在中国被开发的模型”。being强调开发处于进行之中,developed表示模型承受开发这一动作。可复用结构:名词 + being + 过去分词。
- are open source说明模型具有“开源”属性:open source在这里放在are之后描述模型的性质:其相关内容可被公众获取和使用。它不是are opening之类的动作。可复用结构:主语 + be + 性质描述。
- meaning解释前面整个判断:meaning引出的内容说明“这些模型是开源的”具体意味着什么,而不是只修饰China或source。可理解为which means that。
- built与made是共享are的并列被动动作:they指前面的AI models。模型是两个动作的承受者:先被构建,再被免费公开。第二个made前省去了重复的are;被动表达把重点放在模型经历了什么,而不是谁执行了动作。
- for anyone...to...指出谁能够下载使用:anyone是download和use的实际执行者,with the proper computer equipment进一步限定为“拥有适当电脑设备的人”。download和use共享to,也共享未明说的宾语AI models。
核心词汇
- asserting more power /əˈsɜːtɪŋ mɔː ˈpaʊə/ verb phrase
- 扩大或强化权力、管控权
- claiming or exercising greater authority or control
- assert power over · assert greater control · assert one's authority
- breaching /ˈbriːtʃɪŋ/ present participle
- 侵入;突破安全防线
- breaking through or violating a protected system or boundary
- breach security · breach a system · breach confidentiality
- change of tone /ˌtʃeɪndʒ əv ˈtəʊn/ noun phrase
- 态度或论调的转变
- a noticeable shift in the attitude or style being expressed
- mark a change of tone · signal a change of tone · noticeable change of tone
- escalation /ˌeskəˈleɪʃən/ noun
- 升级;加剧
- an increase in the seriousness, intensity or extent of something
- escalation of tensions · rapid escalation · prevent further escalation
- taken responsibility /ˌteɪkən rɪˌspɒnsəˈbɪləti/ verb phrase
- 承担责任
- accepted accountability for an action, event or outcome
- take responsibility for · accept responsibility · assume full responsibility
- intervened /ˌɪntəˈviːnd/ verb
- 介入;干预
- became involved in a situation in order to influence what happened
- intervene in a dispute · intervene to prevent · directly intervene
- industrial-scale /ɪnˈdʌstriəl skeɪl/ adjective
- 达到产业规模的;大规模的
- occurring on a very large or systematically organized scale
- industrial-scale production · industrial-scale theft · industrial-scale operation
- sanctions /ˈsæŋkʃənz/ plural noun
- 制裁
- official penalties imposed to pressure a country, organization or company
- impose sanctions · face sanctions · lift sanctions
原文来源:BBC News · Trump considering AI controls after OpenAI hacking incidents(版权归原出版方所有,本站仅提供中文讲解)