Post #1625876

2026-04-21 14:45 UTC

I know there is a lot of hype and panic surrounding #anthropic and #mythos but I think the thing I am looking forward to and what no one seems to focus on is that 1) the flaws in code are still there Mythos or not, and 2) imagine running this as a part of a CI pipeline to look at every bit of code particularly during a &quot;build&quot; and &quot;test&quot; stage. That second part is what I am really looking forward to. #security #infosec #ai

Replies (3)

@ryanjyoder@techhub.social 2026-04-21 14:53

@simplenomad Talk about flakey tests. 😂 "How did you fix that security warning?" "Oh I just rerun the CI when it fails."

Open ##1822965
@mathaetaes@infosec.exchange 2026-04-21 17:05

@simplenomad the real question is. Is that truly realistic. Without independent, objective evaluation from someone who isn’t already an anthropic fanboy or PR press release, we don’t know what the limitations or tradeoffs are. How many tokens were burned per finding? How much time did it take? How much user interaction is required? I’m also looking forward to being able to do “automated pentesting” as a pre-release step in addition to our own human testing, but the way this was announced and handled thereafter makes it really hard to tell if it’s truly amazing or just cherry-picked cases to hype up the next round of funding and/or an IPO.

Open ##1822969
@Mutesplash@uncontrollablegas.com 2026-04-21 17:18

@simplenomad I imagine that being very expensive and requiring lots of manual intervention.

Open ##1822970