4 Matching Annotations
  1. Last 7 days
    1. We reviewed a demonstration of this specific technique being used to identify a small number of previously known, minor vulnerabilities. These vulnerabilities all appear relatively simple, and we have found that other publicly-available models are able to discover them as well without requiring a bypass.

      这是一个重要的技术声明,质疑政府行动的合理性。Anthropic声称发现的漏洞是已知的、微小的,且其他模型也能发现。这需要独立验证,以确定政府反应是否过度,以及Fable 5的安全性是否真的如Anthropic所描述的那样。

    2. Our understanding is that the government believes it has become aware of a method of bypassing, or 'jailbreaking' Fable 5. We reviewed a demonstration of this specific technique being used to identify a small number of previously known, minor vulnerabilities.

      这里包含了需要核实的技术细节。Anthropic声称政府发现的'越狱'方法仅能识别一些已知的、次要的漏洞,且其他公开模型也能发现这些漏洞。需要独立验证这一技术评估的真实性和准确性,以及政府所关注的安全问题的严重程度。

  2. May 2020
    1. after nearly 10 years of continuous improvement

      Not necessarily a good or favorable thing. It might actually be preferable to pick a younger software product that doesn't have the baggage of previous architectural decisions to slow them down. Newer projects can benefit from both (1) the mistakes of previously-originated projects and (2) the knowledge of what technologies/paradigms are popular today; they may therefore be more agile and better able to create something that fits with the current state of the art, as opposite to the state of the art from 10 years ago (which, as we all know, was much different: before the popularity of GraphQL, React, headless CMS, for example).

      Older projects may have more technical debt and have more legacy technologies/paradigms/integrations/decisions that they now have the burden of supporting.