OpenAI could watermark ChatGPT-generated text, but here’s why it may not just yet

Today, ChatGPT can do pretty much everything from building itineraries, drafting poems to generating images. But with its growing capabilities, there comes the glaring need to differentiate between AI and human content.

To tackle this, several platforms are coming up with subtle ways to label AI-generated content. OpenAI too has now announced its plans to watermark ChatGPT-generated text, but it may not deploy it to users just yet. Here’s why.

OpenAI AI text detector: What's there to know?

OpenAI mentions how the text watermarking feature was originally planned for students resorting to ChatGPT-generated text for their assignments or exams. According to the company, the feature would essentially employ algorithms capable of embedding subtle text markers, which will eventually highlight ChatGPT-generated text.

The markers will not be invisible to the naked eye; however, the tool will be able to tell its origin by studying a specific format of words and phrases. Despite its obvious benefits, OpenAI may not be looking at rolling it out anytime soon as it might deter those using the model for benign purposes.

In a statement provided to TechCrunch, OpenAI mentioned how it is taking a ‘deliberate approach’ due to the various complexities involved that could eventually impact the broader ecosystem.

ALSO READ: OpenAI announces a major ChatGPT privacy update; here’s what you should know

The tool, which takes a different approach to detecting AI-generated text, is reportedly more powerful than other existing AI-detectors today. If that’s not all, the company also stated that the tool will only be able to detect ChatGPT-generated text and will not study other models.

OpenAI AI text detector: When is the tool likely to roll out?

The tool is said to be currently under development, and owing to its larger impact, OpenAI may not be planning to roll it out anytime soon. That said, it is certainly in the pipeline, as OpenAI recognises it as a boon to the industry.

Watermarking, in the long run, could play a critical role in ensuring transparency and tackling misinformation in the field of content creation. While it is hard to tell if or when the tool could be ready for use, the move is definitely a positive advancement for those dealing with AI on a day-to-day basis.