Interesting work. Stopping AIs from getting accurate results from stolen Knowledge Graphs using RAG. I wonder how long, until the AI companies find a workaround…
But I think it is a move in the right direction for protecting IP. One of the reasons my employer is so slow to look at AI is losing their IP. If the KGs can be poisoned, so that internal systems work with the data, but if the KG is stolen, the LLM RAGging it will produce spurious or misleading information, that is a huge win for industry and R&D labs, for example.