LogFAQs > #971762501

There are talks of watermarking chatgpt outputs via encoding certain phrases in certain ways, so that will stop the college cheaters. It won't stop the scammers with the dedication to fork the code and remove the waterstamps.

There are videos on it, but the tl dr is that text produced from these mainstream engines will always be detectable as AI output. That will slow down the school cheaters, since kids won't find the obscure patched AI engines.

So Amazon books will likely use undetectable(ish) AI as there's a profit incentive to put those up.