Thread - Nostr Hypermedia

If you let AI analyse code that's known to be safe, it will still find something to change and weaken the security over time.

Security Degradation in Iterative AI Code Generation:A Systematic Analysis of the Paradox

So I don't understand what OpenSats is trying to achieve here,..

↑ Parent

Replies (1)

LightningBuck 3 weeks ago

Paper is from 19 May 2025. Ai has improved a lot since then. Also the paper is about AI changing the code, in the case oft the red Team the AI is only used to find the vulnerability which is then disclosed to the development team. So hopefully it's humans implementing the changes.