That’s something that might need to be revisited, unfortunately. Pulling a HG title and issuing refunds is pretty damaging to the brand, and I’m sure is something the company’s not keen to do again.
With that said, I don’t have an answer for how to catch AI-gen HG material in future. As we’ve talked about on different threads, some of the things people confidently declare as signs of GPT authorship (overuse of “embarked” or “amidst,” or excessively vague/purple prose more generally) will throw a lot of false positives. As @Lan noted, if you have enough of an author’s writing handy to spot a huge contrast between things they’ve clearly written themselves and the prose of most of the game text, it can be a giveaway. That’ll catch ESL writers a lot more often than native speakers, though–for the latter, the quality gap will often be more questionable, and pressing into it will be more witch hunt-y. And if nearly the whole game was written using AI, beta testers won’t always have much grounds for comparison.
As Lady Luck said, HG has never screened for prose quality above a super-basic threshold, and doesn’t have the institutions in place to do so. But something new is going to be needed to prevent disasters like this… and I don’t think a “don’t ask don’t tell” policy about AI use is going to cut it.