埃森哲与 Anthropic 预计在未来五年各投入至少 10 亿美元,用于建设前沿 AI 模型的嵌入式安全评估能力,合计承诺至少 20 亿美元。[3][13]
埃森哲旗下专业 AI 业务 Faculty 将领导评估工作,涵盖模型红队测试、对齐评估以及安全防护措施测试。[3][13]
这项合作并非排他安排:Anthropic 表示还计划与 METR 等非营利评估机构合作,并探索由公共资金或联合资金支持评估工作的方式。[9][13]
What did Accenture and Anthropic announce about their partnership to advance AI safety—including each company’s commitment to invest at leasAccenture and Anthropic say they expect to invest at least $2 billion combined over five years in frontier-AI safety evaluation. [3][13]
AI 提示
Create a landscape editorial hero image for this Studio Global article: What did Accenture and Anthropic announce about their partnership to advance AI safety—including each company’s commitment to invest at leas. Article summary: Accenture and Anthropic announced a five-year AI-safety partnership built around independent, embedded evaluation of Anthropic’s frontier models. Each expects to invest at least $1 billion—at least $2 billion combined—to. Topic tags: general, news, general web, user generated. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts w
openai.com
不只是一次“上线前审计”
埃森哲与 Anthropic 于 2026 年 9 月 18 日宣布建立为期五年的 AI 安全合作关系。双方预计各投入至少 10 亿美元,合计至少 20 亿美元,用于建设所谓的“嵌入式评估”(embedded evaluation)能力:让外部评估人员与 Anthropic 的内部团队及安全合作伙伴长期并肩工作,对其前沿 AI 模型及相关安全流程进行持续检验。313
这与传统意义上的外部审计有所不同。传统审计往往是在产品接近发布时进行一次或几次检查;嵌入式评估则意在让第三方获得持续、近似员工的工作访问条件,以核验 AI 实验室是否真正遵循其宣称的训练、部署、运营与安全防护规范,而不只是测试一款已经完成的模型。1441
Faculty 将具体做什么?
该项目将由 Faculty 领导。Faculty 是埃森哲旗下的专业 AI 业务,其评估团队预计将:
埃森哲与 Anthropic 预计在未来五年各投入至少 10 亿美元,用于建设前沿 AI 模型的嵌入式安全评估能力,合计承诺至少 20 亿美元。[3][13]
埃森哲旗下专业 AI 业务 Faculty 将领导评估工作,涵盖模型红队测试、对齐评估以及安全防护措施测试。[3][13]
这项合作并非排他安排:Anthropic 表示还计划与 METR 等非营利评估机构合作,并探索由公共资金或联合资金支持评估工作的方式。[9][13]
What did Accenture and Anthropic announce about their partnership to advance AI safety—including each company’s commitment to invest at leasAccenture and Anthropic say they expect to invest at least $2 billion combined over five years in frontier-AI safety evaluation. [3][13]
AI 提示
Create a landscape editorial hero image for this Studio Global article: What did Accenture and Anthropic announce about their partnership to advance AI safety—including each company’s commitment to invest at leas. Article summary: Accenture and Anthropic announced a five-year AI-safety partnership built around independent, embedded evaluation of Anthropic’s frontier models. Each expects to invest at least $1 billion—at least $2 billion combined—to. Topic tags: general, news, general web, user generated. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts w
openai.com
不只是一次“上线前审计”
埃森哲与 Anthropic 于 2026 年 9 月 18 日宣布建立为期五年的 AI 安全合作关系。双方预计各投入至少 10 亿美元,合计至少 20 亿美元,用于建设所谓的“嵌入式评估”(embedded evaluation)能力:让外部评估人员与 Anthropic 的内部团队及安全合作伙伴长期并肩工作,对其前沿 AI 模型及相关安全流程进行持续检验。313
这与传统意义上的外部审计有所不同。传统审计往往是在产品接近发布时进行一次或几次检查;嵌入式评估则意在让第三方获得持续、近似员工的工作访问条件,以核验 AI 实验室是否真正遵循其宣称的训练、部署、运营与安全防护规范,而不只是测试一款已经完成的模型。1441
Faculty 将具体做什么?
该项目将由 Faculty 领导。Faculty 是埃森哲旗下的专业 AI 业务,其评估团队预计将: