I think the issue is that the security issue is inherent to the core story being told about AI. “It can read your emails and respond automatically” is both a pretty compelling story to certain people who aren’t waist deep in tech news, and also exceedingly dangerous.