近日，微软软件工程部门经理 Shane Jones 发现 OpenAI 旗下的 DALL-E 3 模型存在一处漏洞，该漏洞能够生成一系列 NSFW（不宜内容）的不当内容。Shane Jones 在上报这一漏洞后，却遭到了 OpenAI 的“封口令”。尽管如此，他还是选择将这一情况公之于众。
DALL-E 3 是 OpenAI 的一款图像生成模型，其能力强大，可以基于文本描述生成相应的图像。然而，此次漏洞的暴露显示出即使在顶尖的人工智能模型中，也存在潜在的风险。这不仅对 OpenAI 本身，也对使用其技术的企业和个体提出了警示。
在人工智能技术日益发展的今天，保障其安全性、避免滥用和不当应用变得越来越重要。此次漏洞的曝光也将促使 OpenAI 和其他相关企业更加重视产品的安全性测试和监管。
Title: OpenAI’s DALL-E 3 Model Exposes Pornography-Generating Bug
Keywords: DALL-E 3, Inappropriate Content, Bug
News Content:
Recently, Microsoft software engineering manager Shane Jones discovered a bug in OpenAI’s DALL-E 3 model, which is capable of generating a series of NSFW (inappropriate content) images. After reporting the bug, Shane Jones was issued a “gag order” by OpenAI, but he still chose to disclose the issue to the public.
DALL-E 3 is an image generation model developed by OpenAI, which possesses powerful capabilities and can generate images based on text descriptions. However, the exposed bug raises concerns about potential risks even in top-tier AI models. This not only applies to OpenAI but also to businesses and individuals using its technology.
As artificial intelligence technology continues to advance, ensuring its security, preventing abuse, and avoiding inappropriate applications become increasingly important. The exposure of this bug will urge OpenAI and other related companies to pay more attention to product security testing and supervision.
【来源】https://www.ithome.com/0/748/569.htm
Views: 2