I suspect we’ll only see more and more vulnerabilities like this exposed as LLMs are used by capable security researchers and computer criminals to reduce the amount of work required to explore concepts that wouldn’t have been worthwhile in the past.

The article doesn't mention anything about LLMs. Neither does the code repository or the paper , with the exception of

This paper was edited for grammar using Grammarly and Google Gemini.

in the Acknowledgements.

What makes you think that an LLM made this work possible? The technique of using an interposer to mess with memory writes is both an obvious one and one that has been used for previous attacks, as mentioned in the article.