OpenAI has developed a watermarking tool designed to identify text generated by ChatGPT, but its release has been delayed due to internal debates and concerns about its impact on non-native English speakers.

This ongoing debate over the tool's implementation has persisted for two years, with proponents arguing that it would help combat misuse of AI in academic settings.

While some OpenAI employees advocate for the tool's release, others express concerns about potential user distrust and the risk of false accusations of cheating.

The new detection technology boasts an impressive accuracy rate of 99.9%, although it does not work with other language models and can be circumvented through translation or modification.

Public opinion appears to favor AI detection tools, with a recent survey indicating a strong four-to-one majority in support of their development.

Despite the promise of AI detection tools, experts caution against relying solely on them for accusations of academic dishonesty.

The tool is seen as potentially beneficial for educational institutions, helping educators identify AI-generated assignments amid rising concerns about academic integrity since the launch of ChatGPT.

OpenAI is actively researching effective methods for identifying AI-generated text while ensuring that user trust in their tools is not compromised.

Concerns about the tool's potential impact on user retention and transparency have influenced OpenAI's decision-making process regarding its release.

The debate continues, with proponents arguing for responsible action in releasing the tool, while opponents fear it could negatively affect OpenAI's profits.

Internal conflicts have hindered the deployment of the watermarking tool, reflecting OpenAI's cautious approach to ensure its effectiveness and reliability.