# Anthropic 公布独立评估机制细节,METR 团队获类员工访问权限
> 原标题:An update on independent evaluators
- 来源:Anthropic:Newsroom
- 发布时间:2026-10-14T16:00:00.000Z
- AIHOT:https://aihot.news/items/xl51s0hyyzi1krwrmown4hoqh
- 原文:https://www.anthropic.com/news/an-update-on-independent-evaluators
## 摘要
Anthropic 宣布已向评估前沿 AI 模型的非营利机构 METR 的六名员工提供类员工访问权限,包括公司硬件、办公空间和与内部风险评估团队相当的内部系统访问,METR 将公开发布风险评估结果,Anthropic 不得否决发布但可要求删减少数预定义类别的信息。
## 推荐理由
原文由 Anthropic 亲自说明独立评估机制的具体安排,读者可以了解第三方如何获得接近员工的访问权限和发布权。
## 正文 · 中文译文
上个月,我们承诺对我们的模型和安全实践进行[更强有力的独立第三方评估](https://darioamodei.com/post/we-must-pace-the-frontier)。我们还签署了《超级智能白宫协议》,其中各公司承诺与独立的外部审计机构或评估机构合作。今天,我们宣布已开始向来自[METR](https://metr.org/)的团队提供类似员工级别的访问权限,METR是一家评估前沿AI模型的非营利组织。
METR将公布其风险评估结果。在过去一年中,我们与METR合作进行了[多项针对我们风险报告的审查](https://metr.org/risk-assessment/)。我们还在与[SecureBio](https://securebio.org/)和[Accenture](https://www.anthropic.com/news/accenture-embedded-evaluation)作为评估方合作,并正在与包括[Hugging Face](https://huggingface.co/)在内的其他几家组织进行早期讨论。
在此,我们解释与评估方合作的方式。
## 我们为何与独立评估方合作
随着模型变得更加强大,它们需要更强的保障措施和更具挑战性的风险评估。我们希望对自己的决策公开负责。但我们无法公开分享与安全工作相关的所有信息。其中一些信息可能对恶意行为者有用,或者属于商业敏感信息,例如算力或训练信息。因此,我们让第三方评估方深度访问我们的信息,并让他们分享关于构建我们模型所涉风险的证据和有理有据的结论。我们旨在缩小评估方对前沿AI公司系统的了解与公司自身了解之间的差距。
要评估前沿模型,评估方需要具备以下条件:
- 在评估前沿模型能力方面的_专业知识_,包括理解评估可能出错的方式,例如[对齐伪装](https://www.anthropic.com/research/alignment-faking)。
- _独立性_:他们不应持有被评估公司的股权,也不应在财务或运营上依赖该公司。
- 足够的_访问权限_,以便有意义地评估风险。除了访问最新模型之外,评估方还需要了解这些模型是如何训练的、我们的保障措施如何运作等信息。
- _自由发布_对模型或流程的观察和评估 。
尽管这一类比并不完全准确,但银行监管的运作逻辑类似。银行比外部人士更了解自身的风险状况,因此监管机构会向最大的银行派驻常驻检查员。如果银行的风险承担可能威胁公众,第三方会试图提前发现。
对于AI行业而言,这是迈向[稳健安全监督](https://www-cdn.anthropic.com/files/4zrzovbb/website/0a58d567024a8b448ff15158ebc3625328dfcc1f.pdf)的一步。我们正在与多个评估方快速推进,并积极了解何种合作方式行之有效。我们承诺会持续分享最新进展。
## 我们所说的“类似员工级别的访问权限”是什么意思
我们已为六名METR员工完成入职,为他们提供公司硬件、办公空间,以及对内部工具和系统的访问权限,其权限大致与我们自己的风险评估团队所拥有的权限相当。除少数明确定义的类别(例如客户数据、受律师-客户特权保护的信息,或有关我们模型架构的信息)之外,Anthropic员工被鼓励像对待从事对齐风险工作的同事一样与他们分享信息。1
METR 将发布其对我们风险水平、事件或实践的分析。Anthropic 不能否决发布,但如果信息属于上述等狭窄的预定义类别,我们可以要求他们进行删改。METR 可以自由地公开描述我们的删改情况,并说明其是否认为删改是合理的,只要不泄露被删改的信息即可。
我们与 METR 的接触仅限于一个在可执行协议下运作的小团队。METR 的其他员工与该团队的工作相互隔离,包括与我们合作其他项目的研究人员。嵌入式评估让少数在 Anthropic 安全标准下工作的合格人员能够近距离了解我们的运作方式。这并不是其他组织获取我们或我们客户及合作伙伴信息的方式。
## 我们的其他评估方合作
我们正在推进若干其他合作。由于这些合作仍处于早期阶段,我们向这些评估方提供的信息较为有限,他们的发布权也较为有限。我们在这些早期合作中的目标是确保合作伙伴保护机密信息,并具备评估我们模型所需的资源和专业知识。随着合作推进和信任建立,我们希望充分嵌入更多独立评估方。
我们正在为评估方制定三个层级的访问权限。随着时间推移,我们希望有多个评估方达到第三层级,每个团队都能发布自己对我们风险态势和关键决策的分析。
**在第一层级,**评估方开展范围狭窄的风险调查。我们决定向第一层级评估方提供何种访问权限,评估方可在我们批准后发布其发现。
**在第二层级,**评估方无需我们批准即可发布,并且可以访问更多信息。我们仍可要求移除特定类别的信息,但评估方可就我们的删改是否合理发表看法。
**第三层级**是我们与 METR 所描述的安排,具有类似员工的访问权限,以及类似于第二层级的发布权。
**SecureBio** 目前处于第一层级。在完成并[发布了对我们上一份风险报告的试点审查](https://securebio.org/resources/anthropic_feb_2026_risk_review.pdf)后,它正在审查我们[8月风险报告](https://www-cdn.anthropic.com/f61d49fa5596956a5dec75fea0e973bf6a6a8378/Redacted%20Risk%20Report%20August%202026%20.pdf)中的化学和生物武器章节。审查完成后,我们将考虑转向具有更多访问权限和发布权的项目。
**与 Accenture** 和 Faculty,我们正在[推进一项雄心勃勃的合作](https://www.anthropic.com/news/accenture-embedded-evaluation),包括评估和对模型进行红队测试、开展对齐评估、测试模型防护措施,以及在我们各组织内建设能力。
我们是自愿嵌入评估方的,我们认为每一家前沿 AI 公司都应这样做。
## 相关内容
### 为2026年美国中期选举做准备
今年早些时候,我们分享了选举防护措施的更新。在此,我们概述了在今年美国中期选举前我们如何继续加强这些防护措施。
[阅读更多](https://www.anthropic.com/news/2026-us-midterms)
### Anthropic 开设新加坡办公室并与新加坡政府签署谅解备忘录
Anthropic 已在新加坡开设办公室,并与该国政府签署谅解备忘录,以推进安全、负责任地采用 AI。
[阅读更多](https://www.anthropic.com/news/singapore-office-opening)
### 2026 使用政策更新
我们正在发布新版使用政策。在本文中,我们总结了所做的更改。
[阅读更多](https://www.anthropic.com/news/2026-usage-policy-update)
## 正文 · 原文
Last month, we committed to [stronger independent, third-party assessment](https://darioamodei.com/post/we-must-pace-the-frontier) of our models and safety practices. We also signed the White House Accord on Super Intelligence, in which companies pledged to partner with independent external auditors or evaluators. Today, we’re announcing that we’ve started to give employee-like access to a team from [METR](https://metr.org/), a nonprofit that evaluates frontier AI models.
METR will publish the results of its risk assessments. Over the past year, we’ve worked with METR on [several reviews of our Risk Reports](https://metr.org/risk-assessment/). We’re also working with [SecureBio](https://securebio.org/) and [Accenture](https://www.anthropic.com/news/accenture-embedded-evaluation) as evaluators, and are in early discussions with several other organizations, including [Hugging Face](https://huggingface.co/).
Here, we explain our approach to working with evaluators.
## Why we work with independent evaluators
As models have become more powerful, they’ve required stronger safeguards and more challenging risk assessments. We want to be publicly accountable for the decisions we’re making. But we can’t publicly share all the information relevant to our safety work. Some of it could be useful to malicious actors, or is commercially sensitive, like compute or training information. Instead, we’re giving third-party evaluators deep access to our information, and are having them share evidence and reasoned conclusions about the risks involved in building our models. We aim to close the gap between what evaluators understand about a frontier AI company’s systems and what the company itself understands.
To evaluate frontier models, evaluators need the following:
- _Expertise_ in assessing the capabilities of frontier models, including understanding the way evaluations can go wrong, such as [alignment faking](https://www.anthropic.com/research/alignment-faking).
- _Independence_: they shouldn’t hold equity in the company they’re evaluating, or be financially or operationally dependent on it.
- Enough _access_ to meaningfully evaluate risks. Beyond access to the latest models, evaluators also need visibility into how those models were trained, about how our safeguards function, and more.
- _Freedom to publish_ observations and assessments of a model or a process.
Although the analogy isn’t exact, bank supervision operates on similar logic. Banks know more about their own risk positions than outsiders do, so regulators assign resident examiners to work inside the largest banks. If a bank’s risk-taking could threaten the public, third parties try to see that in advance.
For the AI industry, this is a step toward [robust safety oversight](https://www-cdn.anthropic.com/files/4zrzovbb/website/0a58d567024a8b448ff15158ebc3625328dfcc1f.pdf). We’re moving quickly with multiple evaluators and learning actively about what kinds of partnership work. We’re committed to sharing updates as we go.
## What we mean by “employee-like access”
We’ve onboarded six METR employees with company hardware, office space, and access to internal tools and systems that are roughly on par with what our own risk assessment teams have access to. Outside a small number of clearly defined categories (e.g., data about customers, attorney-client-privileged information, or information about our model architectures), Anthropic employees are encouraged to share information with them just as they would with a fellow employee working on alignment risk.1
METR will publish its analysis of our risk levels, incidents, or practices. Anthropic can’t veto publication, but we can require them to redact information if it falls into narrow, pre-defined categories like the ones above. METR is free to characterize our redactions publicly, and say whether it believes the redactions were justified, as long as it does so without revealing the redacted information.
Our engagement with METR is limited to a small team operating under enforceable agreements. Other METR employees are siloed from that team’s work, including researchers who work with us on other projects. Embedded evaluation gives a few qualified people working under Anthropic’s security standards a close view of how we operate. It’s not a way for other organizations to obtain our information, or that of our customers and partners.
## Our other evaluator partnerships
We’re working on several other partnerships. Because these other partnerships are still in early stages, we’re providing these evaluators with more limited information, and they have more limited publication rights. Our goal with these early engagements is to make sure that our partners protect confidential information and have the requisite resources and expertise to evaluate our models. As partnerships progress and we build trust, we hope to fully embed more independent evaluators.
We’re developing three tiers of access for evaluators. Over time, we hope to have several evaluators at the third tier, with each team publishing their own analysis of our risk posture and key decisions.
**In the first tier,** evaluators conduct narrowly scoped risk investigations. We decide what access to provide a first-tier evaluator, and evaluators can publish their findings with our approval.
**In the second tier,** evaluators can publish without our approval, and they have access to more information. We can still require removal of specific categories of information, though evaluators can publish their views on whether our redactions were justified.
**The third tier** is the arrangement we’ve described with METR, with employee-like access and publication rights similar to those of the second tier.
**SecureBio** is currently in the first tier. It’s in the process of reviewing the chemical and biological weapons chapter of our [August Risk Report](https://www-cdn.anthropic.com/f61d49fa5596956a5dec75fea0e973bf6a6a8378/Redacted%20Risk%20Report%20August%202026%20.pdf), after completing and [publishing a pilot review](https://securebio.org/resources/anthropic_feb_2026_risk_review.pdf) of our previous risk report. After this review is complete, we will consider moving to projects with increased access and publication rights.
**With Accenture** and Faculty, we're working [toward an ambitious partnership](https://www.anthropic.com/news/accenture-embedded-evaluation) that includes evaluating and red-teaming models, conducting alignment assessments, testing model safeguards, and building capabilities across our organizations.
We’re embedding evaluators voluntarily, and we believe every frontier AI company should do the same.
## Related content
### Preparing for the 2026 US midterms
Earlier this year, we shared an update on our election safeguards. Here, we outline how we’ve continued to strengthen these safeguards ahead of this year’s US midterms.
[Read more](https://www.anthropic.com/news/2026-us-midterms)
### Anthropic opens Singapore office and signs MOU with the Government of Singapore
Anthropic has opened an office in Singapore and signed an MOU with the country’s government to advance safe, responsible AI adoption.
[Read more](https://www.anthropic.com/news/singapore-office-opening)
### 2026 Usage Policy update
We’re publishing a new version of our Usage Policy. In this post, we summarize the changes we’ve made.
[Read more](https://www.anthropic.com/news/2026-usage-policy-update)