It's hard not to see the development of LLMs as a major stepping stone towards a much more hostile web. Downstream from there, the open web will likely be a wasteland and all interactions will be suspect. There seems like a world where long-running, personalized phishing campaigns could be a reality too.