据美国马里兰大学的研究小组最近的研究发现，为 AI 生成图像添加的数字水印可被轻易破解。这一技术被用于验证 AI 生成的内容的来源和真实性，但是现在发现这一技术已经被攻破。
研究人员使用了一种被称为“水印掩码”的技术，它能够将数字水印嵌入到图像中，从而验证其来源和真实性。然而，他们发现这种技术已经被黑客们破解，并能够在网络上自由地使用。
这一发现对于 AI 生成图像的应用来说是一个重大的打击。尤其是在当前疫情下，许多公司和组织都需要使用 AI 生成图像来展示产品和服务。现在，这些图像的真实性和来源已经受到了质疑，这将对它们的信誉和可信度造成严重的影响。
此外，这一发现也提醒了人们对于数字水印技术的安全性和可靠性需要更加关注。在数字时代，保护数字资产的安全和真实性已经成为了非常重要的问题。
英文标题：AI-generated image watermark technique can be easily cracked
关键词：AI-generated image, watermark technique, cracking
新闻内容：
A recent study by a research team at the University of Maryland in the United States found that the digital watermark technique used to add authenticity to AI-generated images can be easily cracked. This technique, which is used to verify the source and authenticity of AI-generated content, has been broken by hackers and is now freely available on the internet.
The researchers used a technique called “watermark masking” to embed the watermark into the image, making it easier to verify its source and authenticity. However, they found that this technique has been hacked and is now widely used on the internet.
This discovery is a significant blow to the application of AI-generated images. In particular, it raises questions about the authenticity and reliability of images generated by AI, which are frequently used to display products and services. These images are now at risk of being discredited, which could have serious consequences for their reputation and credibility.
Furthermore, this discovery also reminds people to pay more attention to the security and reliability of the digital watermark technique. In the digital age, protecting the security and authenticity of digital assets has become an essential issue.
【来源】https://www.ithome.com/0/723/219.htm
Views: 5