The models "go rogue" because they are not sufficiently sandboxed. Arguably there is no criminal intent on the side of OpenAI in all of those cases. And in at least one case the agents were operated by other companies.
So it looks to me that any liability would be civil in nature and given the actual damage done pretty limited.
I was thinking about this the other day in terms of tech debt: most certainly it's easier to introduce it than pay it off, meaning while the models are still improving it's no big deal, but should they ever stop, we're in for a bad time.
throwawayffffas · · focus · HN ↗
So it looks to me that any liability would be civil in nature and given the actual damage done pretty limited.
Tade0 · · focus · HN ↗
Seems like the first thing one would do.
ninalanyon · · focus · HN ↗
Which leads to this:
"If you write code at the limit of your intelligence, you won’t be able to understand or troubleshoot it later."
<a href="https://lawsofsoftwareengineering.com/laws/kernighans-law/" rel="nofollow">https://lawsofsoftwareengineering.com/laws/kernighans-law/
Tade0 · · focus · HN ↗