Conversation

Notices

1. Embed this notice
   Albert ARIBAUD ✎ (aaribaud@mastodon.art)'s status on Friday, 24-Apr-2026 18:07:57 JST Albert ARIBAUD ✎

   in reply to

   • Tristan Nitot✓

   @nitot

   Je me demande quelle est la probabilité que de plus en plus de projets soient produits à l'aide de l'IA en ignorant consciencieusement cette question de la maintenabilité humaine. 😬

   • Embed this notice
     Tristan Nitot✓ (nitot@framapiaf.org)'s status on Friday, 24-Apr-2026 18:07:59 JST Tristan Nitot✓

     in reply to

     " It is complex, but not arbitrarily complex(1).

     The defects are finite, and we are entering a world where we can finally find them all.

     (1) There’s a risk that codebases begin to surpass human comprehension as a result of more AI in the development process, scaling bug complexity along with (or perhaps faster than) discovery capability. Human-comprehensibility is an essential property to maintain, especially in critical software like browsers and operating systems."

   • Embed this notice
     Tristan Nitot✓ (nitot@framapiaf.org)'s status on Friday, 24-Apr-2026 18:08:00 JST Tristan Nitot✓

     in reply to

     (3/4, oui je sais j'avais dit /2 au départ...) "Encouragingly, we also haven’t seen any bugs that couldn’t have been found by an elite human researcher. Some commentators predict that future AI models will unearth entirely new forms of vulnerabilities that defy our current comprehension, but we don’t think so. Software like Firefox is designed in a modular way for humans to be able to reason about its correctness. It is complex, but not arbitrarily complex"

   • Embed this notice
     Tristan Nitot✓ (nitot@framapiaf.org)'s status on Friday, 24-Apr-2026 18:08:02 JST Tristan Nitot✓

     in reply to

     (2/2) "We have many years of experience picking apart the work of the world’s best security researchers, and Mythos Preview is every bit as capable. So far we’ve found no category or complexity of vulnerability that humans can find that this model can’t."

   • Embed this notice
     Tristan Nitot✓ (nitot@framapiaf.org)'s status on Friday, 24-Apr-2026 18:08:04 JST Tristan Nitot✓

     Article intéressant par Mozilla sur le recours à l'IA pour trouver des trous de sécurité - https://blog.mozilla.org/en/privacy-security/ai-security-zero-day-vulnerabilities/
     "Elite security researchers find bugs that fuzzers can’t largely by reasoning through the source code. This is effective, but time-consuming and bottlenecked on scarce human expertise. Computers were completely incapable of doing this a few months ago, and now they excel at it. " (1/2)