[freedmand]

Do you think that there are security or data exfiltration risks with their approach, e.g. putting classified documents/emails into private company's LLMs to summarize information or identify certain ideological motivations? I don't see how they could begin to understand the scope of what they're working with without A.I. assistance.