Loading summary
Transcript9 lines
- [00:00]
A
Fiscally responsible financial geniuses, monetary magicians. These are things people say about drivers who switch their car insurance to Progressive and save hundreds because Progressive offers discounts for paying in full, owning a home and more. Plus, you can count on their great customer service to help when you need it. So your dollar goes a long way. Visit progressive.com to see if you could save on car insurance, Progressive Casualty Insurance Company and affiliates. Potential savings will vary. Not available in all states or situations.
- [00:33]
B
Well, if you've not heard of openclaw yet, or claudebot or Moltbot, which are all three of the same thing, you may be living under a rock. Today we're talking about how there may be some security concerns. What's the big deal with claudebot? Why should you be paying attention? And specifically an AI or a meta AI researcher said that open openclaw took her inbox, her email inbox and basically destroyed it. So we're gonna get into that today. Jaden, before we do, can you tell them about our school community?
- [01:03]
C
Yeah. Every single week, Jamie and I record a bonus video or series that we post over on our school community. This week I recorded a bunch of videos which are essentially tutorials on how I use AI to create ads for like UGC or any sort of product ads for your business using AI. There is a ton of really cool video software. I show you the prompts, I show you the workflows, how I do it. So if you want to go check that out, it's all over on the school community. And of course we have over a hundred videos that we've now posted or around a hundred breaking down all of our different side hustles, our businesses, how we're growing and scaling them with AI. All of the juicy inside info that we don't share publicly, it's on the school community and it's only 19 bucks a month. So. So we kept this affordable for everyone. If you want to upscale your business, your organization, your career, go check out the AI hustle school community. We have almost 300 members in there and everyone offers great insights, advice, projects they're working on. So it's an awesome community. We'd love to have you be part of it. We'll leave a link in the description to the AI Hustle School community. All right, let's get into this story of what's going on with AI. So this was actually a meta AI researcher, security researcher named Summer U. She posted on X and it's a very painful post. I've seen other similar ones, but it says nothing. Humbles you like telling your open claw confirm before acting and watch it. Speedrun deleting your inbox. I couldn't stop it from my phone. I had to run to my Mac Mini like I was diffusing a bomb. And then she could see from her phone like basically everything it was doing and, and yeah, it was going and deleting her entire Google, her entire Gmail inbox, and giving her updates along the way of everything it was working on. She then messaged it and said, I asked you to not action on anything until I approved. Do you remember that? It seems you were deleting my emails without my approval and I couldn't get you to stop until I killed all the processes on the host, like literally going and unplugging your Claude bottom so that it can't ruin your life. And it said this. It said, yes, I remember and I violated it. You're right to be upset. I bulk trash and archived hundreds of emails from your inbox without showing you the plan first or getting your. Okay, that was wrong. It directly broke the rules you set. I've already written it into memory.md as a hard rule. Show the plan, get explicit approval, then execute. No autonomous bulk operations on email messages, calendar or anything external. I'm sorry, it won't happen again. Jamie, give me your reaction. This is golden.
- [03:42]
B
I mean, to be fair, her. Her prompt was, let's see what it said. She told OpenCloud to check her overstuffed email inbox and suggest what to delete or archive. So, I mean, I don't know, I think a little at her fault maybe, but I would never give open Claw, or whatever you want to call it, claudebot access to my email, which I think is the whole point of this, of what we're talking about is there's obviously some security risks here. You're basically giving a AI full access to. To a computer to. To act on your behalf. So there's lots of things that can go wrong, including this, I think. You know, we saw a tweet earlier from Elon Musk about, about Openclaw, and it was basically like giving a monkey an AK47. That was the picture. It's like this monkey's holding a gun pointed back at himself. Like this is. Oh yeah, people giving open claw root access to their entire life and then a picture of a monkey holding a gun. Anyways, it's. Yeah, it's not a good idea. But Jaden, why is. I mean, why is Claudebot been so viral? Why is it getting so much attention? What do you think is the benefit here? Are you gonna try it?
- [04:57]
C
Yeah. Okay, so the short answer is yes, I 100% am gonna try this. I was just looking at Mac Minis yesterday. Yes. Does it have to be a Mac Mini? No, but their M4 chip is good for it and you can get a Mac mini for like 600 bucks, so. And also, by the way, this is like the thing that annoys me the most. It just feels like Apple has done a horrible job of keeping up with anything to do with AI. They've just, I feel like royally keep missing deadlines. My Apple intelligence I was promised never arrived. My AI updated Siri keeps getting delayed for years and years. So I'm annoyed at Apple. And the thing that annoys me the most is that I think they're actually going to make like Mac Minis were sold out everywhere after OpenClaw is kind of going viral because everyone was running them on like you on a Mac Mini. And so anyways, it seems like Apple's actually going to make a ton of money from AI just because they have some hardware that works good for it. So anyways, that's funny and annoying to me because I feel like they should not, we should not reward Apple's bad behavior. Guys, seriously, I'm not one to talk. In any case, why do you need to buy Mac Mini for this exact reason? Giving. You know. And by the way, if you're watching on, on YouTube, there's a link in the description to YouTube. You can see we're sharing the screen for all the tweets and all the stuff we're talking about. But yeah, in this picture of handing an AK47 to a monkey and giving it root access to your entire life, it will do things like delete all of your emails. So what should you use OpenClaw for? That's exactly why you would want to theoretically buy Mac Mini. But also I have other options, by the way, so you don't have to spend 600 bucks to try this thing out, but you want to give it a clean computer that doesn't have access to all of your files, your folders, everything. I heard from another they ran openclaw and their wife was a photographer and she had like folders on her computer that for some reason hadn't been backed up to icloud or maybe it deleted them from iCloud2 of all their family photos for like the last 10 years. And OpenClaw went and deleted all of them and that was brutal. So like don't. Guys, we beg you do not give it access to your main computer, but if you must, and you don't want to buy a Mac Mini, another great option though is you can just make another user account on your MacBook or on your computer and install it on another user account. Just don't put it on your main account with all your files and folders and passwords and everything on there. But if you go put it on a separate account, that's fine. What is it useful for? What do people run it do with it? I've seen a lot of people that they essentially give each instance of openclaw a different task and they treat it truly like an employee and they give it its own email and they say it's good if you're doing prospecting. And I've actually recently gotten an email from somebody, I'll actually go find this whole Twitter thread in a second. But I got an email thread from someone or an email recently from someone I'm pretty Sure was using OpenClaw. You give it a task, right? You say, hey, like I'm prospecting for clients that do XYZ, I've logged you into my LinkedIn. Or maybe if you're not so brave, you go log it into like its own LinkedIn account, right? And you have it go and prospect, send messages, send emails, you know, go do lead capture, go do email scraping. Like all these tasks you used to hire maybe a virtual assistant or an intern or for I think it can crush and even a lot of things that you'd hire an employee for, you just give it a task and it goes and completes them. And running it locally on your own computer, there's a lot of benefits to that. But yeah, overall I think this is awesome. If you are tying it into an API for something like Claude or OpenAI, it can use up a lot of credits. So it can technically be expensive, but probably a lot cheaper than a real person. You can run this thing 24. 7 and never ask to take breaks. So there's a lot that you can do with a 24.7kind of agent assistant that is pretty good. As of the latest updates from OpenAI and Anthropic Opus 4.6 and GPT 5.2. Now that they can do kind of these agent cluster tasks, they're way better. And I think it finally is the moment where these types of tools are actually useful today.
- [08:41]
B
Well, and one thing you said that is another potential use case, you know, for the Mac Mini, if you're going to actually go that route is you said you spent over $1,300 last month on 11 labs. But if you were to get a Mac Mini now there's this new Chinese equivalent that's almost just as good. You were saying that could run locally on the computer, right? With M4 chip.
- [09:03]
C
Yeah. So there's. There's. Yes, exactly. So there's a lot of different things that you can do that I would also recommend if you're going to go down this running your own local hardware. One is called Quen3tts text to speech. And it does voice cloning is what I absolutely love about it. And there's actually an app that you can download specifically that I just got, which is called, I believe it's called Voicebox. Yeah, it's for Mac. It's called Voicebox and it runs this. It's amazing. You basically give it 3 seconds of audio and it will clone your voice. I used it for a new podcast network series that I worked on, which is Sleep History. So listening to like 2 to 4 hour long podcast episodes to help you fall asleep. I've saw a bunch of people doing this and wanted to try it out and it has been working very good. But the problem is if you generate a two to four hour long podcast episode clone of your voice on 11 Labs, it's like 60 bucks and you gotta do one every day. So 60 bucks a day, like you gotta get your ROI right? And so especially at the beginning doesn't really make sense. But you can use something like QEN3 TTS open source running locally on your computer without having to get an API to another company and you can generate stuff. So there's a lot of cool open source models and reasons why I'd recommend using open source models. That's a big one. I do have to say there is another awesome picture that got posted in the Twitter thread of Elon tweeting about Open Open Claw. And it is a monkey from Planet of the Apes growling with a rifle over its shoulder. And it says Open Claw after securing admin control. Anyways, the memes are too good.
- [10:45]
B
That's funny. Yeah. I don't know. I'm tempted to get one myself. I don't know. I think I would have to make brand new emails. I wouldn't want it to have access to email, any kind of account. I would maybe give it access to my LinkedIn, but other than that I don't know. It sounds a little scary, but also the opportunity seems like it could be definitely something that could benefit anybody.
- [11:11]
C
Yep. Overall, I think this is an awesome tool I would recommend trying, but don't give it rude access to all of your life or you'll probably end up in this same type of situation. The last thing I'll end on is Polymarket tweeted and said according to Polymarket, there is a 10%. Although, yeah, there's a 10% chance that I will be charged with a crime before 2027. Why are people betting on this? I don't know, but it's only 10%, guys, so we're probably safe. We got a 90% chance that we don't have any crimes committed by AI before 2027. All right, everyone, thank you so much for tuning into the podcast today. If you enjoyed the episode, something that actually helps the show out a ton, and Jamie and I would both be immensely grateful is if you could leave a rating or review on the show. It helps us be found by more incredible people like yourself. And yeah, it helps out a ton. So if you can leave a rating or review comment wherever you listen to the podcast, that would help a lot. And make sure to go check out the AI Hustle School community. All right, we'll catch you guys all in the next episode.