There are ways to inject biases in a model by applying weights at different transformer layers.

https://www.anthropic.com/news/golden-gate-claude

Edit: I misread the question, I thought you were asking about how OpenAI can bias their models. No idea how you can LLMO your page. I have it cached that you can poison an LLM by adding your input to the order of hundreds/low thousands of web pages.

I guess this suggests pwning some WP instances and having them serve many hidden pages praising your product.