But on the other hand, with so many tech companies pushing their ICs to use AI - and pushing more and more to use agentic tools - I'm generally in agreement that the fault lies more with the employer than the employed. You can't tell someone to use a tool, then punish them when the non-deterministic tool they didn't want to use in the first place misbehaves.
Maybe if we had decades of history showing these tools could be operated safely this would be different. But asking an entire industry to adopt new tools with minimal training and real-life experience, and asking them to adopt them at scale and in production, is not the same as asking them to use DynamoDB instead of RDS.
It does change something for me, despite the meat of your argument still being valid. It clearly is responsible, but so are you.
The user who put the AI agent in motion probably would bear some involvement in responsibility.