“The watermark looks completely natural to those reading the text because the choice of words is mimicking the randomness of all the other words.”
– Search engine marketer
Roger Montti, describing forthcoming cryptographic watermarks that will make texts generated by language models like ChatGPT instantly detectable. Summarizing research computer scientist
Scott Aaronson is doing for
OpenAI, Montti considers both the utility and fallibility of the security measure.
