The article makes it clear that (as the author understands it, at least) someone who uses open source software in their commercial product is liable; the people who wrote the open source code [1] are not.
> If a user is harmed by software, the person they paid (targeted ads would count) must compensate them for the harm – unless the software provider can prove their software played no role in the ... harm. If open source resources are [used by] your code, you’re responsible for their performance too. *The open source resource licensed away their liability to you*.
(Emphasis mine)
[1] Assuming they used a license that limits liability, such as Apache.
The article is attempting to create a scare about things that have always been true. If a telco's services crash the telco has to compensate customers even if it was a postgres failure that caused it by failing to authorise handsets for a connection in a cell. For example.
The telco has service agreement with customers and it's clear exactly what service it was supposed to do and failed. Where is such agreement for a random github repository? To put it a bit ad absurdum, say user supplies parameter to your math function so that it divides by zero and it results in some injury or loss. Who is liable for that? Shold judge try to parse some piece of code for whether it was reasonable for user to expect passing zero will work?