The major social media sites know absolutely that their products expose customers and users to predictable harm and cause real damages. They should bear an analogous liability burden as a business operating a physical location.
Sounds great. Last I checked, most businesses don't have a legal obligation to police the conversations that happen on or around their premises.
You're also wrong though - a corporation is absolutely liable for threats and various types of verbal harassment that occurs within their walls. Thus the endless liability-offsetting mandatory training videos... Duty of care is well established.
From its employees or contractors, possibly. From customers or visitors? Absolutely not.
> ...the conversation is an integral part of the product.
For telephone, text messaging, and mail [0] the conversation is an integral part of the product. And yet, somehow they don't have this "liability burden" you're asserting is borne by everyone who provides a service or a space where people could plausibly speak with each other.
[0] ...both e- and snail- variants...
https://www.mbm-law.net/insights/premises-liability-for-thir...
Here's a more nuanced example discussing liability to foreseeable harm by a third party on a business's premises:
https://www.supremecourt.ohio.gov/rod/docs/pdf/0/1995/1995-O...
It is of course differently illegal to use the mail (in the US at least) for criminal activity.
Another way to think about this is, if this liability didn't generally exist, why do platform providers require a special protection?