Claude Code codebase leaked
Anthropic: ‘human error, not a security breach’
https://www.youtube.com/watch?v=OpkyF-ugVq4&list=UU9rJrMVgcXTfa8xuMnbhAEA - video
https://pivottoai.libsyn.com/20260401-claude-code-codebase-leaked - podcast
time: 8 min 19 sec
https://pivot-to-ai.com/2026/04/01/claude-code-codebase-is-leaked/ - blog post
I'm genuinely impressed. Claude Code is somehow worse than I ever could have imagined. This is incredible.
That’s the relentless march of progress for you doing what corporate America does best and improving with every day
Never used it.
Am I right?
Not gonna lie here, when code like this is leaked, I think it is a cry from help from inside.
Someone couldn’t take it anymore
@moses_izumi aahahahahahaa lol
er, no
@davidgerard "our vibecoded software has leaked its code, that's human error" should be yet another cautionary tale to anyone still thinking their bosses aren't using them as an accountability sink. Every time I hear a working-class person parrot "of course you have to double-check the AI output" I want to scream.
(that said, I have very little sympathy for Anthropic employees in this particular case)
The "undercover" mode is interesting.
If Anthropic doesn't want people to know they themselves are using Claude to contribute to repos, why should the rest of us use it?
@davidgerard love the observation that vibe coded slop can't be copyrighted, so the DMCA orders they're putting out are not only wrong, they're fraudulent
Though if the prompts were written by humans they can possibly claim copyright over those I suppose
> Remember: every day is AI Fool’s Day.
geeeero
David Gerard
gaytabase
Jacob Alexander Tice
Moses Izumi
lit
Mark Whybird
Matt Palmer
GhostOnTheHalfShell
Wulfy—Speaker to the machines
ndevenish
Lucas
Everyday Freethought
Jer
Janik 
Errant 
