To answer this question, one should first understand how the AI generates content.
A common misconception is that the AI goes out into the internet, lift relevant content from online sources, and rewrites it. This is wrong.
In actual fact, the AI is based on a generative language model. In other words, it predicts the next few words, given a sequence of words, with a certain amount of variability or randomness. The AI does not actually "understand" what it writes.
If you truly comprehend how the AI works, you would also understand that plagiarism, in general, is not an issue.
The chances of your generated article being identical to another is practically zero. Sure, there might be the occasional instance where a phrase or two appears somewhere else (especially if it's a common phrase), but this happens by chance, rather than by "plagiarism". It's highly unlikely that Google would penalize a site just because one or two phrases appear in another site.
If you are still worried about "plagiarism", you could always use a third-party plagiarism checker. There are both paid and free plagiarism checkers available online. A note of caution: different plagiarism checkers define "plagiarism" differently, so you are likely to get different results depending on which tool you use. Regardless, all plagiarism checkers in the market differ from Google's own definition of plagiarism (which is ultimately what matters), but that definition is not public.
GPT-3 does not guarantee that its generated content is free from "plagiarism", however that is being defined. That goes the same for WordHero.