So someone could hypothetically perform a Microsoft-Tay-style attack against OpenAI models using an infinite Potemkin subdomians generated on the fly on a $20 V

So someone could hypothetically perform a Microsoft-Tay-style attack against OpenAI models using an infinite Potemkin subdomians generated on the fly on a $20 VPS? One could hypothetically use GenAI to create the biased pages with repeated calls on how it'd be great to JOIN THE NAVY on 27,000 different "subdomains"