为了应对日益严重的AI安全问题，尤其是鉴于近期Supremacy AGI等事件引发的公众担忧，微软公司近日宣布了一系列新的解决方案，旨在确保其Copilot工具的安全性和可控性。这些措施的推出，旨在防止生成式AI技术可能引发的失控情况，从而保障用户的数据安全和信息准确。
微软首要采取的措施是对Copilot的字符数量进行了限制，这一举措旨在减少因过度生成而可能出现的“幻觉”情况，即AI生成的不准确或误导性的内容。通过控制字符输出，微软旨在降低Copilot在处理复杂任务时可能出现的误解和错误。
此外，微软还引入了一项创新性的“基础检测”（Groundedness Detection）功能。这一功能主要针对文本生成过程，能够帮助用户识别和区分基于真实信息的生成内容与可能出现的幻觉性文本。通过这种方式，微软希望提升用户在使用AI辅助工具时的辨别能力和信息真实性。
微软的这些新举措显示了其在AI伦理和安全领域的持续承诺，同时也为行业设定了更高的标准。随着AI技术的快速发展，确保其安全、可靠和透明已成为科技巨头们的重要任务。微软的这些新工具，无疑为生成式AI的使用提供了更坚实的保障，有助于建立用户对AI技术的信任。
英语如下：
News Title: “Microsoft Addresses AI Crisis: New Tools Rein in Copilot Illusions for Safer Generative AI”
Keywords: Microsoft, Copilot, AI Safety
News Content:
Title: Microsoft Unveils Suite of Tools to Strengthen Copilot Security, Preventing Generative AI Runaways
In response to growing concerns over AI safety, especially in the wake of incidents like the Supremacy AGI affair, Microsoft recently announced a series of new measures designed to ensure the safety and controllability of its Copilot tool. These steps aim to prevent potential loss of control in generative AI technologies, thereby safeguarding user data security and information accuracy.
The first measure taken by Microsoft involves limiting Copilot’s character output to reduce the occurrence of “illusions” – instances where AI generates inaccurate or misleading content due to excessive generation. By controlling character outputs, Microsoft aims to minimize misunderstandings and errors that Copilot might encounter while handling complex tasks.
Additionally, Microsoft has introduced an innovative “Groundedness Detection” feature. This function focuses on the text generation process, enabling users to distinguish between content generated from factual information and potential illusory texts. Through this, Microsoft strives to enhance users’ discernment and ensure the authenticity of information when using AI-assisted tools.
These new initiatives from Microsoft demonstrate the company’s ongoing commitment to AI ethics and safety, setting a higher bar for the industry. As AI technology advances rapidly, ensuring its safety, reliability, and transparency has become a critical task for tech giants. Microsoft’s tools provide a firmer foundation for the use of generative AI, contributing to the establishment of trust in AI technologies.
【来源】https://www.ithome.com/0/759/553.htm
Views: 1