奥特曼承认公众有理由担忧AI风险但呼吁信任行业自律OpenAI Chief Altman Says World Right to Fear AI Risks but Urges Trust in Firms

在软件企业Salesforce于旧金山举办的年度会议上，OpenAI首席执行官山姆·奥特曼向与会者表示，随着AI工具的能力快速提升，外界想象这项技术如何失控并不需要像过去那样丰富的想象力，因此世界有充分理由对此感到害怕。

Speaking at an annual conference hosted by software firm Salesforce in San Francisco, OpenAI chief executive Sam Altman told attendees that the rapid advancement in the capabilities of AI tools means it no longer requires as much imagination to picture how the technology could go wrong, acknowledging that the world is right to fear potential failures.

奥特曼同时提出，公众应当对AI企业抱有更多信心，信任各家公司会出于责任感采取正确行动来确保技术安全。他强调，行业内部已经深切感受到了该技术所带来的重大影响与责任分量。这是奥特曼在上周前Anthropic研究员发表有关AI风险的言论引发网络热议后，首次公开发表评论。

Altman argued that the public should place greater faith in artificial intelligence companies to maintain safety standards, asserting that firms will do the right thing because they recognize the true magnitude of the technology. His appearance marked the first time he had spoken publicly since a former Anthropic researcher's viral post regarding AI dangers gained widespread attention last week.

上周，一名从AI企业Anthropic离职的研究人员在网络上发布警告，声称若缺乏有效约束，人工智能可能在本十年末消灭全人类。该言论发布后，多位AI高管和行业专家公开表示认同这一评估，不过相关人士并未详细阐明得出该结论的具体推导过程，也未解释AI系统将如何具体实现这一结果。

Last week, a researcher who resigned from artificial intelligence firm Anthropic posted a warning online claiming that unchecked AI could kill all humans by the end of the decade. A handful of other AI executives and industry experts responded by expressing agreement with the assessment, though they did not detail how they reached that conclusion or explain the exact mechanisms through which AI could cause such an outcome.

在奥特曼发表讲话后，Meta首席执行官马克·扎克伯格在社交平台X上发文指出，每家AI企业都具备采取自主安全措施的能力与内在动力。扎克伯格表示，任何忽视对齐研究的实验室都会在竞争中掉队，同时如果模型造成实际损害，开发机构将面临重大的法律责任风险，因此各企业拥有强烈的动机来预防此类危害发生。

Following Altman's remarks, Meta chief executive Mark Zuckerberg posted on X that every artificial intelligence firm maintains both the ability and the incentive to take its own actions toward safety. Zuckerberg stated that any laboratory failing to focus on alignment will fall behind competitors, adding that labs face significant liability risks if their models cause harm, providing them with a strong motivation to prevent failures.

针对科技高管呼吁信任企业自律的表态，美国政界部分重要人物提出了明确反对，认为不能将如此关键的技术监管责任交由行业内部自行处理。左翼参议员伯尼·桑德斯表示，目前关于人工智能的重大决策权已经被实质上留给了少数全球最富有的富豪手中，缺乏公众监督。

In response to tech executives advocating for industry self-governance, notable US political figures have pushed back against the notion that artificial intelligence firms can be trusted to manage safety autonomously. Left-wing Senator Bernie Sanders criticized the current landscape, asserting that critical decisions regarding AI have been left entirely to a handful of the richest people in the world without adequate public oversight.

与此同时，前美国总统唐纳德·特朗普的顾问兼前白宫策略师史蒂夫·班农也表达了类似的批评意见。班农指出，公众绝不能指望科技寡头实现自我监管。来自左右翼政治光谱两端的知名人物先后发声，共同呼吁政府对快速发展的人工智能行业实施更强有力的外部治理机制。

Echoing those concerns from across the political aisle, Steve Bannon, a former White House strategist and adviser to former US President Donald Trump, stated that the public cannot trust tech oligarchs to regulate themselves. These criticisms from both left-wing and conservative political figures have reinforced growing demands for substantial government intervention and external oversight over the rapidly evolving artificial intelligence sector.

在立法层面，美国国会短期内难以就人工智能安全出台新的法律。尽管议员们在周一结束夏季休会返回华盛顿，但民主党与共和党内部均未就具体监管方式达成共识。随着众议院即将在本周末再次休会以备战11月初的中期选举，实质性立法的推进窗口极为有限。

New federal safety legislation on artificial intelligence remains unlikely to emerge from the US Congress in the near future. Although lawmakers returned from their summer break on Monday, a persistent lack of consensus within both political parties over how to approach regulation continues to stall progress. Legislative momentum faces further timing constraints as the House prepares to enter recess again at the end of the week ahead of midterm elections scheduled for early November.

奥特曼在9月3日接受Axios采访时曾透露，他在2023年出席国会听证会后原本预期议员们会为先进AI建立基础监管框架，但该目标至今未能实现。他指出立法者在跟进技术迭代与避免阻碍创新之间面临两难。与要求设立护栏的声音相反，唐纳德·特朗普批评相关监管提议，将AI安全担忧称为骗局，并主张仅需一位强大而聪明的总统即可充当必要保障。

In an interview published on September 3, Altman stated that he had anticipated Congress would establish a basic framework for advanced AI following his testimony in 2023, though lawmakers have yet to do so. Altman observed that legislators have struggled to keep pace with the rapidly evolving technology while attempting to avoid slowing down innovation. Meanwhile, Donald Trump has rejected calls for guardrails, characterizing fears regarding AI safety as a hoax and asserting that the only necessary protection is a strong and smart president.