OpenAI realises the need to inform people when they see an AI-generated image, especially ones created by its own platform DALL-E 3 and for that the company is offering watermarks. OpenAI has confirmed that it will use the open standard that has been set up by the Coalition for Content Provenance and Authenticity (C2PA). What this means is that any AI image generated using DALL-E 3 will have metadata details like the name of the AI tool used to make the image.
It is an interesting coincidence that OpenAI is bringing its own version of a watermark at a time when Meta has talked about regulating AI-generated content and keeping their identity well and truly clear for the viewer.
And because OpenAI also integrates DALL-E 3 into the ChatGPT for generating AI images, these new metadata details will be coming to the AI chatbot by February 12. Having details about the origin of the image might increase the size of the file, as per OpenAI, but that will have no bearing on the quality of the visual which might be more important to people.
It is good to see OpenAI bring these changes to AI-generated images, however, the company has highlighted a major loophole around the feature that people can easily bypass.
OpenAI has pointed out that social media platforms have a tendency to remove the metadata. You can also lose the details if you take a screenshot of the said image which is quite easy to remove the origins of the image, especially when they are AI-generated.
Situations like these make it evident that stringent regulations for AI content is critical and all tech companies need to work in unison to make it safer to create and consume AI-generated images, videos and more.
Comments
0 comment