According to Odaily, OpenAI has developed a watermark system for texts generated by ChatGPT and has prepared a tool to detect these watermarks. This system has been in development for approximately a year. However, there is internal debate within the company about whether to release the system due to potential impacts on company profits. The company found that when there is a sufficient amount of AI-generated text, this method is '99.9% effective' in detecting AI text. In a survey commissioned by OpenAI, 'one in four respondents globally supported AI detection tools.' Despite this, OpenAI is concerned that the use of watermarks might cause ChatGPT users to lose interest, with nearly 30% of respondents indicating they would reduce their use of the software if watermarks were implemented. Some employees have additional concerns, such as the ease of circumventing watermarks by using techniques like translating text back and forth between different languages with Google Translate or adding and then removing emojis with ChatGPT. Despite these concerns, employees still believe the method is effective. However, given user dissatisfaction, some suggested methods 'might be less controversial among users but remain unproven.'