10 ms·
this seems rather counter-productive, wouldn't a model with less cybersecurity capabilities be more likely to produce insecure code? Not to mention, Chinese mod
by bwat49 3mo ago
this seems rather counter-productive, wouldn't a model with less cybersecurity capabilities be more likely to produce insecure code? Not to mention, Chinese models don't have these restrictions and can be used to exploit said unsecure code.
I supposed I shouldn't be surprised at how the trump admin is approaching AI regulation, counter-productive is really all they do
- ihsw 3mo agoAs contradictory as it sounds, they (Anthropic) are probably trying to dance the fine line where its public models can write secure code but cannot exploit insecure code.