The funny thing about all of this is how LLMs can brute force uncover memory vulnerabilities from C/++ and punish humanity for making the C ABI the target for all software instead of any memory safe language.
Had access at work so no money out of my pocket. Spent $20 actual dollars in about 30 minutes. Still fucked things up in a complex app. Went back to Kimi
it actually helped unfuck our zfs + file permissions that one of our senior engineers was fucking around with the other day and broke our dropbox accessor. it really untangled a web of permissions bullshit that was put in place at different levels for various reasons. first time i’ve ever seen a model fix something that complicated and get it right on the first try. now its gone :(
it’s always a YMMV situation with these tools which is why i don’t think it should cost so much.
our ai is soooo good you’re not even allowed to see it, too dangerous
Part of me thinks this was secretly coordinated to give Anthropic a cover story for capacity/cost issues.
This is so Ed Zitron, and it makes so much sense.
I kind of fell off the series after Fable 3, the first one was the only really good one anyway.
Your health is low. Do you have any potions, or food?
Chicken chaserrrrrr
This is the company that people use to “vibe code”? Seems weird that the US government would shut it down.
Without reading into it at all:
This is either a clap back from anthropic not playing ball
Or
Anthropic’s stupid ass marketing again
I think it’s a combo of both. Trump wants an excuse to punish Anthropic more for refusing to comply and pushing back on military and overt surveillance use, so he uses their heinous marketing against them to paint the model as dangerous
