Having a watermark for AI-generated content could significantly help in its detection, and it was recently revealed by OpenAI that it already has the technology ready for its renowned chatbot, ChatGPT. Despite their development of an AI watermark for its content and a tool that can detect it, the company is still torn between adding the technology or keeping it from appearing.
Other AI companies have already added AI watermarking features for their content, especially with Google having it available for Gemini.
ChatGPT Text Watermark: OpenAI is Torn in Adding It
PAU BARRENA/AFP via Getty Images
OpenAI announced in its most recent blog post that among the latest developments of the company is a new "Text watermarking" feature that is already available from the company. However, according to OpenAI, it is not yet available on its renowned AI chatbot, particularly as the company is still considering its addition and researching "alternatives."
The company claimed that its latest watermarking tool is "highly accurate and effective" against 'localized tampering,' saying that it can fight against those that paraphrase it.
However, 'globalized tampering,' is not yet as effective, particularly for bad actors, using different techniques like translations, using other AI content generators, or adding special characters in between and asking the AI to remove it after.
While one of its reasons for not using it is because the watermarking tool is not yet perfect, OpenAI also found in its recent survey among focus groups that its users may no longer use ChatGPT should watermarks arrive. According to the company, adding watermarks could lead to a stigma where non-native English speakers may opt to use AI as a mainwriting tool.
OpenAI and ChatGPT's AI Detection Tool
Generative AI has proved to the world that it can create content with significant information. Still, not all are reliable, and this also affects OpenAI's ChatGPT, especially for its hallucination tendencies. With that, there have been significant requests to companies regarding detection tools for AI-generated content, and early last year, OpenAI released one for ChatGPT.
However, OpenAI did not stop there, as there are two main platforms from the company that center on generative AI, and the other one is with its AI image creator, DALL-E. Last May, OpenAI introduced a new tool that would be responsible for detecting AI-generated images, particularly for DALL-E, with the company also adding watermarks to its created media.
The world's users and consumers of AI have been at a crossroads regarding the addition of AI detection tools, as well asimproved measures in watermarking content for easier distinctions. OpenAI recently revealed that it already has both of these tools available, but the company is torn between adding this or keeping it away from its platforms, especially with the possibility that users may no longer use it.
