Search
OpenAI to Introduce Invisible Watermarks for ChatGPT Texts in the EU; No Special Characters Required, Effectiveness May Be Compromised Post-Rewriting
2 day ago /
Read about 0 minute
Author：小编
OpenAI is progressively implementing invisible watermarks on texts produced by ChatGPT and Codex within the EU region, adhering to the EU's AI Act stipulations concerning the traceability of AI-generated content. The textGrain text watermarking technology it utilizes does not rely on inserting hidden characters, invisible spaces, or unique punctuation marks. Rather, it subtly modifies word selections during the text generation process by the model, establishing a statistical pattern that remains imperceptible to the human eye yet detectable by machines.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
