Big data processing was a thing before LLM and AI stuff happened. Furthermore, I am sure that current publicly available models are not really optimized in terms of performance, hence I don't agree that it is a threat due to large scale analysis - it was a thing decade ago.