Criminals, bad drivers, and bad passengers probably all use casual carpooling opportunities, yet those continue to operate in many urban areas. With even minimal attention to the problem, any app-based solution will have better opportunity to identify and filter out those problems, so while it's a potential issue, I think it's unlikely to be particularly problematic.

If the app is aiming to making carpooling more accessible, you will see an increase in occurrence of these issues.

The point I'm driving at is that it's still a marketplace and it comes with all the same standard hazards that apply to any marketplace. There are disputes and exceptions that have to be managed. Machines are not good at that yet.

Casual carpooling opportunities don't have enormously wealthy corporations behind them.

Setting the support obligation or morality questions aside for a moment (though I'm on the side of "by facilitating and profiting from this, you accept some responsibility for the quality of that facilitation"), that big difference is still critical: suing the operator of a neighborhood carpool mailing list or somesuch has limited yield. Suing Google has potentially massive yield. So a corporation should think about that sort of risk, just like they would the PR risks, etc, of letting it be an unregulated wild west.

Your position sounds a lot like "light or even largely-community-based self-moderation will be enough" and yet platforms like Twitter and Reddit that have attracted some large controversies and PR shitshows even without real money or real physical harm being involved. Heck, "don't read the comments" is standard internet advice to the point of being cliche.