AI · SEMICONDUCTOR · FACT CHECK
SKD WIREMostly True
OpenAI has developed a watermarking method for text generated by its AI models and plans to apply it first in the European Union, according to reporting on the company's watermarking plans.
OpenAI's watermarking technique embeds detectable patterns into text produced by its models, allowing outputs to be identified as AI-generated. The method is aimed at helping distinguish machine-written content from human-written text.
The initial application of the watermarking is planned for the European Union, where regulatory pressure on AI transparency is strongest. The EU's AI framework imposes disclosure obligations on AI-generated content, making it the leading candidate for first deployment. Details on timing and whether the feature will extend to other regions remain unconfirmed.
OpenAI has long weighed watermarking as one option among several detection approaches, and the company has previously acknowledged internal debate over the impact on certain user groups.
Based on the available evidence, the claim that OpenAI is applying watermarks to AI-generated text with the EU first is Mostly True.