Post #4167716
2026-07-28 06:52 UTC
@mttaggart@infosec.exchange thanks for this extremely thoughtful and illuminating reply.
to answer the one question you posed:
> the means to perform the testing you describe are not novel; they're not secrets. And yet for some reason, it has eluded even mature projects. Why?
because it’s really expensive! it’s expensive to configure, it’s expensive to run all the relevant compute, it’s expensive to triage all the duplicate and dead-end findings.
Replies (1)
-
@glyph@mastodon.social 2026-07-28 06:52
@mttaggart@infosec.exchange “patch the planet” and glasswing are both also quite expensive efforts, taking some of the best engineering talent in the world and pointing them all at security infrastructure problems with the caveat that they can only do it if they use frontier models to do so. this article points at some ways that maybe using these models *is* an efficacious path forward, maybe even the best one at the moment…