Anthropic宣布与咨询巨头Accenture及其AI部门Faculty建立合作，在其内部设立第三方嵌入式评估员1。这一举措将用于评估AI模型、进行红队测试、执行对齐评估和测试模型保护措施1。两家公司计划在未来五年内在该项目上投入至少10亿美元1。Faculty公司由Accenture在1月份收购作为其AI部门1。
该合作体现了Anthropic首席执行官Dario Amodei的计划——在AI实验室内放置第三方安全评估员1。Anthropic在声明中表示："这些评估员不会减少我们的问责，但有助于使其更易于验证。我们模型的安全性仍然是我们的责任。"1此外，Anthropic正在与METR等非营利组织就"试点嵌入式评估"进行对话1，并表示更多评估员将在未来几周内宣布1。消息发布后，Accenture股价在盘后交易中上涨8%1。
Anthropic has announced a partnership with consulting giant Accenture and its artificial intelligence division Faculty to establish an embedded third-party evaluator within Anthropic's operations.1 The evaluator will be responsible for assessing AI models, conducting red team testing, performing alignment evaluations, and testing model safety measures.1 The two companies plan to invest at least $1 billion over the next five years in this initiative.1
Faculty, which Accenture acquired in January as its AI division, will serve as the embedded evaluator.1 This arrangement represents a concrete implementation of Anthropic CEO Dario Amodei's stated plan to place third-party safety assessors inside AI laboratories.1 Anthropic emphasized that these evaluators will not diminish the company's accountability, stating: "These evaluators will not reduce our accountability, but will help make it easier to verify. Our models' safety remains our responsibility."1 The company indicated that additional evaluators will be announced in coming weeks and noted it is in discussions with nonprofit organizations including METR regarding pilot embedded evaluation programs.1 Following the announcement, Accenture's stock rose 8 percent in after-hours trading.1
评论
还没有评论，欢迎留下第一条。