Clawdbot is an exciting new way to spend $300/day on AI with Anthropic.

It's a personal assistant? and it's a shitty security disaster of course

my theory: Anthropic is going for the mobile game model and inducing gambling addiction in the most horrible SV AI bros

https://news.ycombinator.com/item?id=46760237

EDIT: oh lol from reddit:

> "clawdbot" is a crypto currency fueled product with loads of aggresive fans

that explains quite a lot

Clawdbot - open source personal AI assistant | Hacker News

@davidgerard we seriously don't understand why one would want to run code that nobody wrote, let alone feed their entire browser database, app token, and maildir tree into it, but of course people are just jumping at the chance, enthusiastic to do so

@atax1a @davidgerard

My concern is not that nobody wrote the code, it is that nobody looked at it (at least it seems like they never do).

Vibe coders will say "It wrote the tests and they all passed!" How do they know the tests were any good?

Or are they just impressed with large numbers?

@EFreethought @davidgerard nobody wrote the code, nobody can read it. it isn't code, it's the software equivalent of a prion and a gish gallop. the slop overwhelms all reasoning capability. it wouldnt be better if it was read by anyone, that would give it a false sense of legitimacy
@atax1a @EFreethought @davidgerard I swear, you could write a program that just printf()s "Task completed." for a bunch of things, and these rubes wouldn't even question it. You could exfiltrate their home directory, or straight up just delete it, or backdoor their system (for sweet, sweet residential IPs), and they would neither know nor care. Were I evil, I would release "Ultimate AI Helper Bot 8000 McDeluxe", and when you run it, it fills your hard drive with zeros.

@davidgerard it got renamed to moltbot very soon after it being a fucking security disaster got noticed: https://github.com/moltbot/moltbot

very normal behavior for the uhh, checks duckduckgo, ā€œpersonal AI assistant that went viral in a matter of weeksā€ and serves only to burn its users’ money and exfiltrate their data

GitHub - moltbot/moltbot: Your own personal AI assistant. Any OS. Any Platform. The lobster way. šŸ¦ž

Your own personal AI assistant. Any OS. Any Platform. The lobster way. šŸ¦ž - GitHub - moltbot/moltbot: Your own personal AI assistant. Any OS. Any Platform. The lobster way. šŸ¦ž

GitHub

@davidgerard how dare you imply there’s anything the matter with JiaTanBot, the fleet of LLM agents that have access to all of your credentials on all your systems all at once

anyway it’s named cokebot now to reflect our commitment to make it cheaper and less addictive

@zzt see edit, turns out it's a fuckin crypto project looollllll
@davidgerard oh my god every fucking time!
@zzt not sure they are bright enough to be jia tan bot
@davidgerard @zzt given that the moral of xz was "don't trust people negging open source maintainers", the entire industry is jia tanning
@davidgerard @zzt Everything is gambling. Everything is computer.
@theorangetheme @davidgerard @zzt I’ve seen compelling arguments that, here in the U.S. at least, the normalization of gambling via the legalization of sports betting really opened to floodgates to a lot of this brain-pickled bullshit.
@angry_drunk @davidgerard @zzt I would absolutely not be surprised. Societies that abandon care and love and beauty tend to fall prey to seductive garbage.
@zzt @davidgerard I'm going to guess that Anthropic's lawyers sent then a 'yeah, you're not calling it that' letter.
@davidgerard $300/day? Shit, I'll fuck up your code and your calendar and everything else on your computer for half that! (I promise I have substantive things to say, too, but the shitposting is just too cathartic.)
@davidgerard If there are no AI haters, then I am dead.
@theorangetheme working on hating AI after i'm dead too
@davidgerard God: "So like... you know this is eternal bliss, right?"
Me: "I'm BUSY."
@theorangetheme my wife concurs that those two lines are the most me thing we've ever read
@davidgerard I have my middle finger on the pulse of AI loathing. ;)

@theorangetheme @davidgerard

new favorite laptop sticker

@autonomousapps @theorangetheme @davidgerard I want one, is there a way to acquire one?
New Art Print & Sticker: Fuck AI — Geek Calligraphy

dusts off the keyboard Ahem. Well, it’s been a while since we’ve released a product, and this one couldn’t be more timely. Whether the presence of eerily incorrect images on social media bothers you in your soul parts, you cannot abide plagiarism, or you cannot tolerate the environmental damage p

Geek Calligraphy
@ariaflame is that just a watermark on the digital image or does the actual sticker they sell look like that?
@esoterra No idea. This is what the page selling it says https://geekcalligraphy.com/stickers-and-coloring/fuckai
Fuck AI - Sticker — Geek Calligraphy

Display your contempt for ā€œgenerative AIā€* that spits out the wrong number of teeth with this sticker! This image was 100% human generated by a human being. We promise. Round vinyl sticker *Scare quotes used for any and all sarcasm related to ā€˜generative,’ ā€˜artificial,’ and especially ā€˜intellig

Geek Calligraphy
@esoterra But it could just be a watermark for sales purposes. Some of the other stickers have something like it too.
@ariaflame ya, looking at the other stickers on the site, it has to be a digital-only watermark, because it's much worse on some of them.
@esoterra Alas, in the era of people willing to steal your stuff, this is the sort of thing that happens.
New Art Print & Sticker: Fuck AI — Geek Calligraphy

dusts off the keyboard Ahem. Well, it’s been a while since we’ve released a product, and this one couldn’t be more timely. Whether the presence of eerily incorrect images on social media bothers you in your soul parts, you cannot abide plagiarism, or you cannot tolerate the environmental damage p

Geek Calligraphy
@davidgerard @theorangetheme we have a wonderful opportunity for you! DavidGerardBot, trained on your image and memories, only $100/month
@davidgerard I made the mistake of reading the comments and now my brain hurts.
@cthos i just got to the guy who used it to invent the todo / reminder list and now i swallowed my phone so i can simply die in a less painful way
@darby3 I'm having trouble disambigulating what this thing can actually do with what the commenters *think* it can do and/or imagined it doing. Like the one who's like "oh it can orchestrate a browser to do _anything_ trivially"
@cthos Relatedly I just had a weird moment of empathy for someone second hand who I was just told a story of—a non coder sold on the idea of how these things could reduce actual tedium but instead is running into the hallucinatory blah de blah. I feel bad for those who could be sold on this stuff with pure intentions and are finding out the hard way that it all sucks and is bad at what it’s being sold as.

@cthos @davidgerard This thread sent a cold shiver down my spine.

https://news.ycombinator.com/item?id=46760674

I guess at least they know it's insecure? (But don't worry, you can mitigate that with better prompts 😱 .)

What if we will go even further? I have built end-to-end messaging layer for Cla... | Hacker News

@Grouchybeast @davidgerard considering the thing opens up insecured ports right at the start I’m guessing that is not going to end well.
@davidgerard I want to know why they think this banal chat is a good example of what this trash fire can do?!

@akent a bit of endearing racism for the SV tech bros!

EDIT: who are also crypto bros

@davidgerard The ā€œgatcha" model is definitely a part of it. Last few weeks I’ve been up close and personal with people deep into this shit in multiple ways (vibe-coders, chatbot fanciers, etc) and their brains are utterly pickled. I haven't seen so many disorders on display since the last serial killer convention.
@angry_drunk you must understand they've never been so *effective* get with the programme or be left behind
@davidgerard Yeah, Friday I had to listen to three separate dipshit CEOs prattle on about how I was falling behind because I don't spend hours a day letting an auto-complete algorithm tell me I'm a special boy.
@davidgerard no doubt you've seen the posts about it already, but for a laugh (or a cry) just browse through that install script. It tells very little about what it does, but a lot about the people behind it.

@davidgerard Read something about this thing in my Linkedin feed. Guy posting about this seemed like... on crack, such was excited

Someone pointed out about the security nightmare it is. His anwer: couldn't care less as long as I make $$$.

However, since this things read emails, even confidential ones... how long before some very concerned customer finds about this?

@davidgerard for $300 a day you can just get a human assistant