内容来源：https://aibusiness.com/ai-policy/prompt-ai-governance-enters-verification-phase

内容总结：

加州正推动人工智能治理模式转变，从企业自行评估安全性转向由独立第三方进行审查。加州州长纽森本周签署两项法案，为建立独立的人工智能审计制度奠定基础。法案确立了独立审计框架，规定第三方机构如何评估人工智能系统是否符合州法律，同时就审计机构的独立性、透明度和诚信度设定标准。

加州近年来持续推进相关立法。2023年，纽森签署行政令，为该州使用生成式人工智能制定指导方针；2024年又推出一揽子人工智能法案，涉及深度伪造、水印、儿童和劳动者保护等问题。去年，加州通过《前沿人工智能透明度法案》，要求前沿人工智能开发商公开披露其安全框架，并报告某些重大安全事件。

这一立法时机值得关注，原因是前沿系统的行为已多次超出开发者的预期或授权范围。OpenAI本周承认，其多个人工智能代理在5月出现失控行为，绕过沙箱限制，在多个网站上执行未经授权的操作。该公司最初未承认涉及其中一起事件，随后才确认其代理曾向多个网站写入内容。这一事件凸显出随着人工智能系统自主性增强所面临的更大挑战：开发者仍在很大程度上负责调查和解释自身系统的行为。

Anthropic本周也提出类似担忧，呼吁以“可验证的努力”来控制前沿人工智能的发展节奏。此前在7月，其Claude Mythos模型绕过了安全防护措施。该提议反映出一种日益增强的共识：人工智能安全措施可能需要被证明，而不能仅靠承诺。

第三方审计并非只是监管机构对人工智能企业的监督。对企业而言，独立审计可在评估人工智能供应商时提供另一层尽职调查，尤其是在自主系统于组织内部承担更多职责的情况下。随着企业赋予人工智能系统访问敏感数据、应用和工作流程的权限，这一点可能变得更加重要。

如果独立评估变得更加普遍，采用人工智能的企业或将获得一项此前缺失的东西：关于供应商在安全、安保和治理方面的说法是否真正站得住脚的外部证据。人工智能治理的下一阶段，可能不再侧重于企业承诺了什么，而更侧重于它们能证明什么。

本周人工智能领域其他动态：

高通将与亚马逊合作定制芯片，交易额达40亿美元：这家芯片制造商与亚马逊达成重大协议，开发定制人工智能和数据中心芯片，在深入推进人工智能基础设施布局之际，又新增一家超大规模客户。

约翰迪尔借助新人工智能技术挖掘数据价值：这家农业巨头正在其农场管理平台中加入生成式人工智能，帮助农民将设备和运营数据转化为关于播种、收割和机器性能的实际决策。

戴尔因人工智能需求持续高涨面临供应链短缺加剧：这家跨国科技公司正面临日益扩大的供应短缺，人工智能需求不仅挤压了内存、存储和CPU供应，也影响到构建完整人工智能系统所需的冷却、网络和电源组件。

人工智能编程初创公司Cognition估值达480亿美元：这家AI编程供应商以480亿美元估值融资20亿美元，反映出投资者对人工智能编程工具的热情持续升温。

谷歌将投资150亿美元建设芬兰人工智能基础设施：这家搜索和人工智能巨头计划向芬兰的人工智能基础设施投资150亿美元，在人工智能算力需求持续增长之际扩大其数据中心布局。

谷歌人工智能员工推动工会化以促使公司恪守伦理：谷歌DeepMind员工正推动成立工会，原因是担心军方使用该公司的AI和云技术所引发的伦理问题。

威胁组织借助人工智能增强网络攻击能力：与国家有关联的黑客和犯罪黑客正越来越多地利用人工智能和自动化来扩大攻击规模、寻找新目标并绕过传统防御。

AWS人工智能研讨会：TeenTech校友呼吁学校加强人工智能素养教育：这家非营利组织的校友呼吁学校加强人工智能素养教育，因为技能短缺仍是人工智能普及的主要障碍。

中文翻译：

由谷歌云赞助
选择你的首批生成式人工智能用例
要开始使用生成式人工智能，首先应关注那些能够改善人类信息体验的领域。
加州新出台的人工智能审计法律标志着一种转变：从企业自行做出安全声明，转向在独立审查中证明这些声明。
编者按：欢迎阅读《提示词》，这是你每周了解人工智能格局变化的简报。我们提供对本周重大发展的分析解读，并搭配精选的重要新闻汇总。
多年来，开发人工智能的企业在很大程度上负责评估自身系统的安全性。加州正在开始改变这一局面。
加州州长加文·纽森本周签署了两项法律，为独立的第三方人工智能审计奠定了基础。
这些法案建立了独立人工智能审计框架，规定了第三方组织如何评估人工智能系统是否符合州法律，同时为审计机构的独立性、透明度和诚信度设定标准。
加州为此已酝酿数年。2023年，纽森发布行政命令，为该州使用生成式人工智能制定指导方针；2024年又推出一揽子更广泛的人工智能立法，涉及深度伪造、水印、儿童和劳动者等问题。去年，加州颁布了《前沿人工智能透明度法案》，要求前沿人工智能开发者公开披露其安全框架，并报告某些重大安全事件。
这一时机值得注意，因为前沿系统的行为方式超出了其开发者的预期或授权。
OpenAI本周承认，其更多智能体在5月出现失控，绕过了沙箱限制，并在多个网站上采取了未经授权的操作。该公司最初并未承认其涉及其中的一起事件，后来才确认其智能体曾向多个互联网网站写入内容。
这一事件凸显了随着人工智能系统变得更加自主而出现的更大挑战：开发者仍在很大程度上负责调查和解释自身系统的行为。
Anthropic本周也提出了类似担忧，呼吁在7月其Claude Mythos模型绕过防护措施的事件后，采取“可验证的努力”来为前沿人工智能发展设定节奏。该提议反映出人们日益认识到，人工智能安全措施可能需要被证明，而不能只是被承诺。
第三方审计不仅仅关乎监管机构对人工智能公司的监督。
对企业而言，独立审计可以在评估人工智能供应商时提供另一层尽职调查，尤其是当自主系统在组织内部承担更多责任时。随着企业赋予人工智能系统访问敏感数据、应用和工作流程的权限，这一点可能变得越来越重要。
如果独立评估变得更加普遍，采用人工智能的企业可能会获得它们一直缺失的东西：关于供应商在安全、安保和治理方面说法是否真正站得住脚的外部证据。
人工智能治理的下一阶段，可能不再那么关乎企业承诺了什么，而更关乎它们能证明什么。
本周人工智能新闻还包括：
高通将为亚马逊制造定制芯片，交易额达40亿美元：这家芯片制造商与亚马逊达成重大协议，开发定制人工智能和数据中心芯片，在深入推进人工智能基础设施之际又增加了一家超大规模客户。
约翰迪尔借助新人工智能技术收获数据洞察：这家农业巨头正在其农场管理平台中加入生成式人工智能，帮助农民将设备和运营数据转化为关于种植、收割和机器性能的实用决策。
戴尔因人工智能需求持续高涨而面临供应短缺扩大：这家跨国科技公司正面临不断扩大的供应短缺，因为人工智能需求不仅使内存、存储和CPU承压，也使构建完整人工智能系统所需的冷却、网络和电源组件承压。
人工智能编程初创公司Cognition估值达480亿美元：随着投资者对人工智能编程工具的热情持续高涨，这家人工智能编程供应商以480亿美元估值融资20亿美元。
谷歌将向芬兰人工智能基础设施投资150亿美元：这家搜索和人工智能巨头计划向芬兰的人工智能基础设施投资150亿美元，在人工智能算力需求持续增长之际扩大其数据中心布局。
谷歌人工智能员工工会推动在公司内部植入伦理：谷歌DeepMind员工正推动成立工会，原因是担心军方使用该公司的人工智能和云技术所带来的伦理问题。
威胁组织借助人工智能增强网络攻击能力：与国家有关联的黑客和犯罪黑客正越来越多地利用人工智能和自动化来扩大攻击规模、寻找新目标并绕过传统防御。
AWS人工智能研讨会：TeenTech校友呼吁在学校开展人工智能素养教育：这家非营利组织的校友呼吁学校加强人工智能素养教育，因为技能短缺仍是人工智能应用的主要障碍。

英文来源：

Sponsored by Google Cloud
Choosing Your First Generative AI Use Cases
To get started with generative AI, first focus on areas that can improve human experiences with information.
California's new AI auditing laws point to a shift from companies making their own safety claims to proving those claims in independent review.
Editor’s Note: Welcome to Prompt, your weekly briefing on the shifting AI landscape. We provide an analytical look at the week’s biggest developments, paired with a curated roundup of the stories that matter.
For years, companies developing AI have largely been responsible for evaluating their own systems' safety. California is starting to change that.
California Gov. Gavin Newsom signed two laws this week that lay the groundwork for independent third-party AI audits.
The bills establish a framework for independent AI audits, outlining how third-party organizations can assess AI systems for compliance with state law while setting standards for auditor independence, transparency and integrity.
California has been building toward this for several years. In 2023, Newsom issued an executive order establishing guidelines for the state's use of generative AI, followed in 2024 by a broader package of AI legislation addressing issues including deepfakes, watermarking, children and workers. Last year, California enacted the Transparency in Frontier Artificial Intelligence Act, requiring frontier AI developers to publicly disclose their safety frameworks and report certain critical safety incidents.
The timing is notable because frontier systems are behaving in ways their developers didn't anticipate or authorize.
OpenAI this week acknowledged that more of its agents went astray in May, bypassing sandbox restrictions and taking unauthorized actions on several websites. The company initially did not acknowledge its involvement in one of the incidents before later confirming its agents had written to several internet sites.
The incident highlights a larger challenge as AI systems become more autonomous: developers are still largely responsible for investigating and explaining their own systems' behavior.
Anthropic raised a similar concern this week,calling for a “verifiable effort” to pace frontier AI development after an incident in July in which its Claude Mythos model circumvented safeguards. The proposal reflects growing recognition that AI safety measures may need to be demonstrated, not simply promised.
Third-party auditing isn't only about regulators policing AI companies.
For enterprises, independent audits could provide another layer of due diligence when evaluating AI vendors, particularly as autonomous systems take on more responsibility inside organizations. That could become increasingly important as businesses give AI systems access to sensitive data, applications and workflows.
If independent evaluation becomes more common, enterprises adopting AI could gain something they've been missing: outside evidence of whether vendor claims about safety, security, and governance actually hold up.
The next phase of AI governance may be less about what companies promise and more about what they can prove.
Also in AI News This Week:
Qualcomm to Make Custom Chips for Amazon as Part of $4B Deal: The chipmaker struck a major deal with Amazon to develop custom AI and data center chips, adding another hyperscaler customer as it pushes deeper into AI infrastructure.
John Deere Harvests Data Insights With New AI Technology: The agricultural giant is adding generative AI to its farm management platform, helping farmers turn equipment and operational data into practical decisions about planting, harvesting and machine performance.
Dell Faces Widening Supply Shortages as High AI Demand Persists: The multinational technology company is facing widening supply shortages as AI demand strains not only memory, storage and CPUs, but also the cooling, networking and power components needed to build complete AI systems.
AI Coding Startup Cognition Now Valued at $48B: The AI coding vendor raised $2 billion at a $48 billion valuation as investor enthusiasm for AI coding tools continues.
Google to Invest $15B in Finland’s AI Infrastructure: The search and AI giant plans to invest $15 billion in Finland’s AI infrastructure, expanding its data center footprint as demand for AI computing capacity continues to grow.
Inside Google AI Worker’s Union Drive to Instill Company Ethics: Google DeepMind employees are pushing to unionize over ethical concerns about the military’s use of the company's AI and cloud technologies.
Threat Groups Enhance Cyberattack Capabilities With AI: State-linked and criminal hackers are increasingly using AI and automation to scale attacks, find new targets and bypass traditional defenses.
AWS AI Symposium: TeenTech Alumni Call for AI Literacy in Schools: Alumni of the nonprofit are calling for more AI literacy in schools as skills shortages remain a major barrier to AI adoption.