We don't, although I don't think anyone would be opposed to it. Instead, after experiencing a (non-AI-based) hostage scam, our practice is even easier. We call the purported victim directly to confirm or deny the situation.

Makes sense, and seems easy enough to set up except for the barrier of people being uncomfortable talking about serious potentialities.

In basic military training we used code words / challenge words at entry points as one half of two-factor authentication: I see your ID photo ~matches your face, and I hear your response, thus increasing the chances you are who you appear & claim to be.