OpenAI旗下的DALL-E 3模型被曝存在生成“不当内容”的漏洞。微软软件工程部门经理Shane Jones在发现该问题后，向OpenAI报告了这一情况，但随后却遭到了“封口令”的限制。尽管如此，Shane Jones最终还是选择公开这一安全隐患。据悉，DALL-E 3模型能够生成一系列不适宜在工作场合显示（NSFW）的内容，这些内容可能包含色情、暴力或其他敏感信息。此次漏洞的披露引发了对AI模型内容审核机制的质疑，以及对AI开发者与用户之间透明度和责任归属的讨论。
Title: Unveiling of OpenAI DALL-E 3 Content Vulnerability
Keywords: AI Vulnerability, Inappropriate Content, Gag Order
News content:
OpenAI’s DALL-E 3 model has been reported to have a vulnerability that allows it to generate inappropriate content. Shane Jones, a manager from Microsoft’s software engineering department, discovered the issue and reported it to OpenAI, only to face a “gag order” afterwards. Despite this, Jones decided to disclose the security flaw. It is said that the DALL-E 3 model can produce a series of Not Safe For Work (NSFW) content, which may include pornography
【来源】https://www.ithome.com/0/748/569.htm
Views: 1