3 ms·
Surely in return for the value generated by HN'ers pentesting this thing for free, they can provide a hook for HN mods to test strings against its (recent) outp
by jancsika 4y ago
Surely in return for the value generated by HN'ers pentesting this thing for free, they can provide a hook for HN mods to test strings against its (recent) output.
Or Christ, just ask the fucking thing if it matches its recent output and/or style.
If we weren't all living in the shittiest timeline this would already have been solved before the thing was released.
- nonrandomstring 4y agoThese are insightful shortcuts if your threat model is a single instance. ML chatbots have not yet proliferated.