Post

When ChatGPT was new and everyone was anxious to understand the specific pace of AI progress, I figured a significant milestone to watch for was when it could find and exploit brand new software vulnerabilities on command. It's slightly unnerving to see the milestone met today.

Why@why.bsky.world

New Claude found zero-day's in OpenBSD, ffmpeg, Linux and FreeBSD.

red.anthropic.comClaude Mythos Preview \ red.anthropic.com

Macil

@macil.tech

Questions like "Can this replace most programmers?", "Is this smart enough for recursive self-improvement?", and "Is this dangerous enough now to need regulating or stopping?" could be relatively safely dismissed before that milestone.

8/4/26, 4:00 am0 reposts2 likes

Macil

@macil.tech

8 Apr

I don't think the milestone itself being reached means the answers to these questions have flipped right now, but I expect to be less confident on them going forward.