OpenAI DALL-E 3漏洞遭曝光：生成不当内容引争议
近日，微软软件工程部门经理Shane Jones披露了OpenAI旗下的DALL-E 3模型存在漏洞，该漏洞可生成一系列不当内容。
据悉，DALL-E 3是一款强大的图像生成模型，能够根据文本提示创建逼真的图像。然而，Jones发现该模型存在漏洞，能够生成性暗示、暴力或其他不当内容。
Jones表示，他在发现漏洞后向OpenAI报告了这一问题，但OpenAI并未采取有效措施解决。随后，Jones向外界披露了该漏洞，引发了广泛关注。
OpenAI对此回应称，他们已经意识到该漏洞，并正在采取措施修复。同时，他们强调DALL-E 3并非旨在生成不当内容，并敦促用户负责任地使用该模型。
这一事件引发了关于人工智能伦理的讨论。一些专家认为，人工智能模型需要接受严格的审查和监管，以防止其被滥用。而另一些专家则认为，人工智能的快速发展需要更灵活的监管方式。
目前，OpenAI已采取措施修复该漏洞，并加强了对DALL-E 3模型的监控。同时，该公司也呼吁用户在使用该模型时保持谨慎，避免生成不当内容。
该事件也对人工智能行业敲响了警钟，提醒人们在人工智能技术快速发展的同时，必须重视其伦理和负责任使用。
英语如下：
**Headline:** AI Art Tool DALL-E 3 Flaw Raises ConcernsOver Inappropriate Content
**Keywords:** Flaw, Patch, Inappropriate
**News Content:**
A flaw in OpenAI’s DALL-E3 model has been exposed, allowing it to generate inappropriate content, according to Microsoft software engineering manager Shane Jones.
DALL-E 3 is a powerful image generation model that can create realistic images from text prompts. However, Jones discovered that the model could be exploited to produce sexually suggestive, violent, or otherwiseinappropriate content.
Jones said he reported the issue to OpenAI after discovering the flaw, but the company failed to take effective action. He then publicly disclosed the vulnerability, sparking widespread attention.
OpenAI has since acknowledged the flaw and said it is taking steps to fix it. The company stressed that DALL-E 3 is not intended to generate inappropriate content and urged users to use the model responsibly.
The incident has sparked a debate about AI ethics. Some experts argue that AI models need to be subject to strict review and regulation to prevent misuse. Others contend that the rapid pace of AI development requires more flexible regulatory approaches.
OpenAI has now patched the flaw and increased monitoring of the DALL-E 3 model. The company also called on users to exercise caution when using the model to avoid generating inappropriate content.
The incident serves as a wake-up call for the AI industry, highlighting the need to prioritize ethics and responsible use as AI technology rapidly advances.
【来源】https://www.ithome.com/0/748/569.htm
Views: 1