The floodgates are basically held back currently only by openAI's production go-live policies, which are (currently) 1, manual review with an interview; and 2, specifically forbids use cases that can be used for spamming at scale: https://beta.openai.com/docs/usage-guidelines/use-case-guide...

I'm not expecting this to hold for very long (openai has partnered with microsoft; and there are at least 2 groups currently working on replicating it open-source), and expect strong detoriation of overall web content soon afterwards.