[ad_1]
\n<\/p>\n
Following a report from The Wall Street Journal<\/em><\/a> that claims OpenAI has been sitting on a tool that can spot essays written by ChatGPT with a high degree of accuracy, the company has shared a bit of information about its research into text watermarking \u2014 and why it hasn\u2019t released its detection method. According to The Wall Street Journal<\/em>\u2019s report, debate over whether the tool should be released has kept it from seeing the light of day, despite it being \u201cready.\u201d In an update published on Sunday to a May blog post<\/a>, spotted by TechCrunch<\/em><\/a>, OpenAI said, \u201cOur teams have developed a text watermarking method that we continue to consider as we research alternatives.\u201d<\/p>\n The company said watermarking is one of multiple solutions, including classifiers and metadata, that it has looked into as part of \u201cextensive research on the area of text provenance.\u201d According to OpenAI, it \u201chas been highly accurate\u201d in some situations, but doesn\u2019t perform as well when faced with certain forms of tampering, \u201clike using translation systems, rewording with another generative model, or asking the model to insert a special character in between every word and then deleting that character.\u201d And text watermarking could \u201cdisproportionately impact some groups,\u201d OpenAI wrote. \u201cFor example, it could stigmatize use of AI as a useful writing tool for non-native English speakers.\u201d<\/p>\n Per the blog post, OpenAI has been weighing these risks. The company also wrote that it has prioritized the release of authentication tools for audiovisual content. In a statement to TechCrunch<\/em>, an OpenAI spokesperson said the company is taking a \u201cdeliberate approach\u201d to text provenance because of \u201cthe complexities involved and its likely impact on the broader ecosystem beyond OpenAI.\u201d<\/p>\n<\/div>\n