OpenAI Rolls Out Text Watermarks to Meet EU AI Act Mandates
OpenAI has begun a limited rollout of text watermarking and detection technologies, the company said in an article released Monday (Oct. 5). OpenAI built the technologies in response to the European Union’s AI Act, which requires generative artificial intelligence providers to make AI-generated text identifiable in a machine-readable way, according to the article. OpenAI is […] The post OpenAI…
OpenAI has initiated a limited rollout of text watermarking and detection technologies, as disclosed in an article published on Monday (October 5). The company developed these technologies in response to the European Union's AI Act, which mandates the identification of AI-generated text in a machine-readable manner, according to the article.
OpenAI is allowing API customers worldwide to opt-in to text watermarking for selected models, starting from today, with text watermarking disabled by default in the API, the article mentioned. In the coming weeks, OpenAI will introduce an invisible watermark to eligible ChatGPT and Codex text output in the EU, according to the article.
Furthermore, OpenAI is opening applications for access to its text watermark detector, limiting this tool to approved researchers and expert organizations capable of assisting OpenAI in evaluating and refining the technology, the article noted. OpenAI emphasized that text watermarking and detection are still nascent technologies with significant limitations, and opinions regarding their benefits and responsible usage are still evolving.
The company clarified that a watermark indicates that an OpenAI system generated or processed a portion of a passage, not the extent of human editing involved; it does not establish ownership or accountability; it does not link a user to the text; and it does not verify the passage's accuracy, the article explained. Moreover, the lack of a watermark does not confirm that a passage was authored by a human, the article added.
OpenAI stated that they will continue enhancing detection, examining how watermarks endure editing and translation, and investigating methods to differentiate AI assistance from AI authorship more effectively, as per the article. The restricted rollout announced on Monday pertains solely to OpenAI's text provenance tools and does not impact the company's audio and image verification tools, which are publicly accessible, the article mentioned.
Written by urgent.news from PYMNTS's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.