Who is accountable when AI does the work?

"Computers Cannot Make Decisions" drew a sharp reply: if you deleted customer data and blamed Claude, you would still be the one in trouble, so companies clearly know how to hold people accountable for AI actions. Another commenter asks whether the alignment problem was solved if labs "could" stop their models. "LLMs Aren't Inevitable" got a wry comment that it was the first time the reader wished an article were written by AI. In an older thread, an Ask HN proposes a flag for AI-generated articles, citing the guideline that HN is for conversation between humans, and others note AI detectors are unreliable. A separate thread has Cliff Stoll reacting to AI-written content about his own supposed death.

The pattern across these: the argument has moved from "can AI do it" to "who owns the result". Readers want a human name attached to code, prose and decisions.

The nuance is that enforcement is hard. Detection is unreliable, voting can be gamed, and some commenters say the stigma already makes honest LLM-assisted authors feel unwelcome.