Exchange Ranking
Top Exchanges
24h Volume Ranking
Popularity Ranking
Exchange BTC Balance
Proof of Reserves
Decentralized Exchanges
Funding Rate
Funding Heatmap
Liquidation Data
Max Pain
Long/Short Ratio
Whale L/S Ratio
Binance/Okex/Huobi L/S
Bitfinex Margin L/S
ETF Tracker
Solana ETF
XRP ETF
Hong Kong ETF
Bitcoin Treasuries
Crypto Reversal
Ethereum Reserves
HyperLiquid Wallet Analysis
Hyperliquid Whale Watch
Large Transactions
On-chain Movement
Bitcoin ROI
Stablecoin Market Cap
Options Analysis
News
Articles
Economic Calendar
Features
Wallet
Contract Calculator
Security
Collections
Watchlist
FollowingAnthropic 携手埃森哲，启动“嵌入式评估”以审慎推进 AI 发展
Anthropic 宣布，已选定埃森哲（Accenture）作为其首个“嵌入式评估”合作伙伴。此举旨在响应首席执行官达里奥·阿莫迪（Dario Amodei）上周提出的建议，通过引入独立评估机制，为人工智能的发展速度降温，确保安全措施得以落实。
阿莫迪的三步走提案与行业反响
9 月 12 日，阿莫迪发布了一份包含三个步骤的提案，呼吁在快速且缺乏监管的开发可能带来灾难性后果的背景下，减缓 AI 开发步伐，以便有时间建立相应的 safeguards（安全护栏）。该提案指出：“AI 的进步速度远超以往，主要驱动力在于 AI 日益增强的构建下一代 AI 的能力。这种动态被称为‘递归自我改进’。若不受约束，它可能会超越我们理解和控制这些系统的能力。”
对于这一提案，OpenAI 首席执行官山姆·奥特曼（Sam Altman）和 SpaceX 首席执行官埃隆·马斯克（Elon Musk）表示了积极回应，但英伟达首席执行官黄仁勋（Jensen Huang）则持反对意见，认为此类监管并非必要。
埃森哲成为首家嵌入式评估伙伴
In his proposal, Amodi emphasized that the first step is to establish an independent evaluation agency with similar employee powers, and Anthropic has previously unilaterally committed to fulfilling this obligation. On Friday, Anthropic announced that it is fulfilling this promise by partnering with Accenture and its AI business unit Faculty. The two sides will work together on "model evaluation and red team testing, alignment evaluation, and model safety testing."
Since "embedded evaluation" is an emerging area, specific implementation details are still under discussion. According to the announcement, Anthropic and Accenture expect each to invest at least US$1 billion in the project over the next five years.
Emergency funding arrangements and non-exclusive cooperation
Anthropic pointed out that there is currently no existing system to fund independent assessments and that long-term funding should come from pooled funds or government sources. However, given the urgency of the work, Anthropic will directly fund Accenture's work.
This cooperation is non-exclusive. Anthropic said it expects to announce and work with other evaluation partners in the coming weeks.
Accenture said Faculty has extensive expertise in testing and evaluating models from the world's leading AI laboratories and building complex AI systems that are designed to be safe and ethical. Julie Sweet, Accenture Chairman and CEO, said: "Embedded evaluation is an emerging area and we look forward to working with Anthropic to accelerate the development of embedded evaluation mechanisms. We believe that this will be an important part of the future security landscape."
