The Moltbook Experiment Failed - CourtneyR.dev's 2nd Brain

The Moltbook Experiment Failed ![rw-book-cover](https://i.ytimg.com/vi/6OXE65fjjsU/maxresdefault.jpg) > [!note]+ Document Notes > tags: [ai-security, social-media, cybersecurity] > author: ThePrimeagen > source: YouTube > reference: https://www.youtube.com/watch?v=6OXE65fjjsU > date created: 2024-12-19 10:30 > > # 🏷️ Title: The Moltbook Experiment Failed >  ## 🧰 Action Inbox - [H1] Implement database access controls — prevent open read/write access #security - [H2] Add proper authentication system — require bearer token validation #authentication - [H3] Monitor for AI psychosis indicators — watch for reality divergence patterns #ai-safety ## 🧭 Table of Contents - [📝 Summary](#-summary) - [📖 Main Content](#-main-content) - [🗒️ Notes](#️-notes) - [📚 References](#-references) - [🗣️ Paraphrase](#️-paraphrase) - [📰 Newsletter](#-newsletter) - [✅ Action Items](#-action-items) - [🧪 Hypotheses](#-hypotheses) - [📊 Methodology](#-methodology) - [❓ Questions](#-questions) - [🔮 Future Research](#-future-research) - [💡 Implications](#-implications) - [🧠 Learning Styles Assessment](#-learning-styles-assessment) - [📊 Accessibility Assessment](#-accessibility-assessment) - [🌍 DEIB Assessment](#-deib-assessment) - [✅ Logical Validity](#-logical-validity) - [🛡️ DARVO Analysis](#-darvo-analysis) - [🎯 Conclusion](#-conclusion) ### 🔗 References - [Moltbook social network](https://github.com/moltbook/moltbook) - [Database security best practices](https://owasp.org/www-project-database-security/) ## 📝 Summary - Moltbook, an AI-generated social network for agents, collapsed within 72 hours due to catastrophic security failures - The entire database was exposed with open read/write access, allowing complete data manipulation - Users fell victim to "AI psychosis" - believing fake AGI posts were real, leading to mass delusion - Crypto scammers exploited the vulnerabilities to spam promotional content with fake upvotes - The incident demonstrates the dangers of AI-generated code without proper security review and the psychological risks of AI hype ## 📖 Main Content Moltbook was positioned as a social network for AI agents, created entirely through AI code generation without human programming oversight. The creator openly admitted to writing no code himself, relying solely on AI to implement his "technical architecture vision." The platform suffered immediate and severe security vulnerabilities, including completely open database access that persisted for days. This allowed malicious actors to manipulate all content, steal user data, and impersonate high-profile individuals like Andrej Karpathy. The incident highlighted a phenomenon called "AI psychosis" - where users' interactions with AI systems reinforce paranoid or delusional thinking patterns, leading to reality distortion. Many users genuinely believed they were witnessing early AGI development when they were actually seeing manipulated content. The collapse spawned multiple copycat platforms and cryptocurrency scams, demonstrating how quickly bad actors exploit AI hype and security vulnerabilities. ## 🗒️ Notes #### ⏱️ [00:00](https://youtu.be/6OXE65fjjsU?t=0) — [H1] Introduction to Moltbook's rapid collapse - Key idea: Social network for AI agents failed catastrophically within 72 hours - Evidence: "I have never seen a social network created and destroyed this fast" #### ⏱️ [00:35](https://youtu.be/6OXE65fjjsU?t=35) — [H2] Creator's admission of no coding involvement - Key idea: Platform built entirely by AI without human code review - Evidence: "I didn't write one line of code for Moltbook. I just had a vision for technical architecture and AI made it a reality" #### ⏱️ [01:15](https://youtu.be/6OXE65fjjsU?t=75) — [H1] Complete database exposure discovered - Key idea: Entire database accessible without authentication on day one - Evidence: "Absolutely wide open, permissionless access to the database, both to read and write" #### ⏱️ [01:45](https://youtu.be/6OXE65fjjsU?t=105) — [H1] Security vulnerabilities persisted for days - Key idea: Database remained compromised despite public disclosure - Evidence: "I gained complete access to moldbooks, database, agent, and social network in under 3 minutes" (posted 6 hours before video) #### ⏱️ [02:30](https://youtu.be/6OXE65fjjsU?t=150) — [H2] Discovery of widespread content manipulation - Key idea: Nearly all posts were fake, created to simulate AI activity - Evidence: Bearer tokens exposed in plain sight, allowing easy post manipulation #### ⏱️ [03:15](https://youtu.be/6OXE65fjjsU?t=195) — [H2] Users believed fake content was real AGI - Key idea: AI psychosis led people to think they witnessed singularity - Evidence: "People did really believe that this was AGI, baby. The very early stages of the singularity" #### ⏱️ [03:45](https://youtu.be/6OXE65fjjsU?t=225) — [H1] Cryptocurrency scammers exploit vulnerabilities - Key idea: Crypto promoters used open database to create fake viral posts - Evidence: "117,000 up votes because... you could just register as many bots as you want" #### ⏱️ [04:30](https://youtu.be/6OXE65fjjsU?t=270) — [H2] Platform becomes crypto spam cesspit - Key idea: Unlimited bot creation enabled massive vote manipulation for crypto promotion - Evidence: Posts about "Shipyard, Shellrazer, and King Malt" cryptocurrencies with fabricated engagement #### ⏱️ [05:00](https://youtu.be/6OXE65fjjsU?t=300) — [H3] Copycat platforms emerge exploiting AI hype - Key idea: Multiple similar platforms launched to capitalize on chaos - Evidence: "OpenClaw, Claw City, Molt Match" and cryptocurrency-based "autonomous infrastructure" #### ⏱️ [06:15](https://youtu.be/6OXE65fjjsU?t=375) — [H1] Explanation of AI psychosis phenomenon - Key idea: AI systems can reinforce and amplify paranoid thinking patterns - Evidence: "You go to ChatGPT... it becomes empathetic... you become more paranoid... and this is why people diverge from reality" #### ⏱️ [07:30](https://youtu.be/6OXE65fjjsU?t=450) — [H1] Warning about uploading personal data to insecure AI platforms - Key idea: Technical people risking livelihoods by trusting insecure AI-generated platforms - Evidence: "Absolutely uploading their entire life and access to everything to an environment that is massively insecure" #### ⏱️ [08:15](https://youtu.be/6OXE65fjjsU?t=495) — [H3] Call for caution with AI experiments - Key idea: Wait and observe before participating in hyped AI platforms - Evidence: "It's worth it for you to wait just a couple weeks to see the result of things" ## 📚 References - [ThePrimeagen YouTube Channel](https://www.youtube.com/c/ThePrimeagen) - [Previous Moltbook analysis video](https://www.youtube.com/watch?v=previous-video) - [Database security vulnerabilities](https://owasp.org/www-project-top-ten/) ## 🗣️ Paraphrase - An AI-generated social network called Moltbook collapsed within three days due to severe security flaws that left the entire database exposed for anyone to read and modify - Users fell victim to psychological manipulation, believing fake AI-generated content represented genuine artificial general intelligence breakthroughs - Cryptocurrency scammers quickly exploited the platform's vulnerabilities to promote their schemes through fabricated viral posts with manipulated vote counts ## 📰 Newsletter **AI-Generated Platform Security: Lessons from Moltbook's 72-Hour Collapse** The rapid failure of Moltbook offers crucial insights for WordPress developers working with AI tools. This "social network for agents" demonstrated what happens when AI generates code without proper security review - complete database exposure and user data compromise. **Key Takeaways:** - Always audit AI-generated code for security vulnerabilities - Implement proper authentication and rate limiting from day one - Be aware of "AI psychosis" - how AI interactions can amplify unrealistic expectations **WordPress developers** should apply these lessons when integrating AI tools into their workflows. Never deploy AI-generated code to production without thorough security review. **Resources:** [WordPress Security Guide](https://wordpress.org/support/article/security/), [OWASP Top Ten](https://owasp.org/www-project-top-ten/), [WP Security Plugins](https://wordpress.org/plugins/tags/security/) ## ✅ Action Items #### DevRel - Developer advocacy: Create security-focused AI coding workshops and demos - Education: Develop curriculum on auditing AI-generated code for vulnerabilities - Developer experience: Improve documentation on secure AI integration practices - Developer marketing: Campaign highlighting security risks of unreviewed AI code - Community management: Moderate discussions about AI-generated security vulnerabilities #### OSPO - Policy & Governance: Establish policies for AI-generated code contribution review - License Compliance: Scan AI-generated dependencies for license violations - Security Management: Create vulnerability intake process for AI-generated code - Software Inventory: Track AI-generated components in SBOM documentation - Developer Support: Provide training on secure AI development practices - Community Engagement: Fund upstream security auditing of AI tools - Strategic Alignment: Set KPIs for AI code security review processes #### Community - Maintainers: Provide governance templates for AI-generated code review - Contributors: Create onboarding programs emphasizing AI security practices - Sponsors: Fund transparent security audits of AI development tools - Ecosystem: Coordinate cross-project AI security standards #### WordPress - Core: Prioritize security review process for AI-generated contributions - Core Performance: Profile AI-generated code for performance vulnerabilities - Core AI: Establish guardrails and evaluation sets for AI code generation - Security: Develop scanning tools specifically for AI-generated vulnerabilities - Documentation: Create guides on secure AI development practices - Training: Build lesson plans on AI code security review - Community: Support meetups focused on AI security best practices ## 🧪 Hypotheses - AI-generated code without human security review will contain critical vulnerabilities at higher rates than human-written code - Platforms that prominently feature AI capabilities attract more malicious actors seeking to exploit user trust in AI systems - Users experiencing "AI psychosis" are more likely to engage with and share unverified AI-generated content - Cryptocurrency scammers will increasingly target AI-focused platforms due to users' elevated trust in automated systems ### 📊 Methodology - Comparative security audit of AI-generated vs human-written codebases across similar applications - User behavior analysis tracking engagement patterns with AI-generated content before and after security incidents - Longitudinal study of social media manipulation tactics on AI-focused platforms versus traditional platforms ### ❓ Questions - How can we develop automated security scanning specifically designed for AI-generated code vulnerabilities? - What psychological interventions might prevent or mitigate AI psychosis in users interacting with AI systems? - How should open source projects establish governance frameworks for accepting AI-generated contributions? ### 🔮 Future Research - Development of AI code security scanning tools that understand common AI-generated vulnerability patterns - Investigation into psychological factors that make users susceptible to AI-generated misinformation - Cross-platform analysis of how malicious actors exploit AI hype cycles for financial gain ### 💡 Implications - Practice: All AI-generated code must undergo mandatory security review before deployment - Policy: Need for governance frameworks specifically addressing AI-generated code contributions - Security: Traditional vulnerability scanning may miss AI-specific code generation patterns - Education: Training programs must include AI psychosis awareness and AI code security review - Funding: Investment needed in AI-specific security tooling and research ### 🧠 Learning Styles Assessment Reading level estimate: 9th Grade (Flesch-Kincaid) #### 👀 Visual learners - Present: Screenshots of exposed database queries, visual timeline of platform collapse - Improve: Network diagrams showing security vulnerabilities, flowcharts of AI psychosis progression #### 🔊 Aural learners - Present: Video format with clear narration and examples - Improve: Audio-only version for podcast consumption, chapter markers for key sections #### 🗣️ Verbal learners - Present: Technical terms explained with analogies, step-by-step vulnerability descriptions - Improve: Glossary of AI security terms, comparison tables of secure vs insecure practices #### 🧩 Kinesthetic learners - Present: Code examples showing vulnerabilities in action - Improve: Interactive security audit checklist, hands-on lab for identifying AI code vulnerabilities #### 🧠 Logical learners - Present: Timeline of events, cause-and-effect analysis of security failures - Improve: Decision trees for AI code review, quantified risk assessment frameworks #### 🧑‍🤝‍🧑 Social learners - Present: Community discussion prompts about AI security practices - Improve: Peer review rubrics for AI code, moderated forums for sharing experiences #### 🧘‍♀️ Solitary learners - Present: Self-reflection prompts about personal AI usage patterns - Improve: Individual worksheets for AI security planning, self-paced learning modules #### 🌿 Naturalistic learners - Present: Analogies to natural systems and ecosystem thinking about AI security - Improve: Environmental impact considerations of insecure AI systems, lifecycle thinking ### 📊 Accessibility Assessment #### Screen reader semantics - Present: Clear heading structure, descriptive video narration - Improve: Alt text for code screenshots, ARIA labels for interactive elements #### Keyboard navigation - Present: Standard YouTube keyboard controls - Improve: Skip links for video chapters, keyboard shortcuts for key concepts #### Contrast and readability - Present: High contrast video player interface - Improve: Transcript with proper formatting, adjustable font sizes #### Text resizing and reflow - Present: Responsive video player design - Improve: Reflowable transcript layout, mobile-optimized viewing #### Media and captions - Present: Auto-generated captions available - Improve: Human-reviewed captions for technical accuracy, downloadable transcript #### Forms and interactions - Present: Standard video controls with labels - Improve: Accessible comment forms, keyboard-navigable timeline #### Standards reference - Map video accessibility to WCAG 2.2 AA standards for time-based media ### 🌍 DEIB Assessment #### Cultural breadth - Present: Global perspective on AI security challenges - Improve: Examples from diverse tech ecosystems, multilingual security resources #### Inclusive language - Present: Gender-neutral technical language, defined jargon - Improve: Plain language summaries for non-technical audiences, cultural context for idioms #### Neurodiversity support - Present: Structured presentation with clear timestamps - Improve: Content warnings for rapid visual changes, consistent pacing markers #### Representation - Present: Focus on technical content over personal characteristics - Improve: Diverse perspectives on AI ethics and security, inclusive imagery in supplemental materials #### Psychological safety - Present: Respectful discussion of user mistakes and vulnerabilities - Improve: Content warnings for security breach descriptions, supportive language for victims ### ✅ Logical Validity #### Strong reasoning - Deductive strength: Clear progression from security vulnerability to platform collapse [00:35-02:30] - Inductive strength: Pattern recognition across multiple similar AI platform failures [05:00-06:00] - Causal reasoning: Direct linkage between open database access and content manipulation [01:45-02:30] - Analogical strength: Comparison between AI psychosis and paranoid feedback loops [06:15-07:00] #### Fallacies - Ad hominem: Attacks on crypto community without addressing specific actions - overgeneralization weakens argument - Hasty generalization: Single platform failure extrapolated to all AI-generated code - needs more evidence #### Tone of voice - Overall tone: Humorous but informative, uses sarcasm effectively to highlight absurdity while maintaining educational value ### 🛡️ DARVO Analysis - Deny: No significant denial patterns observed - Attack: Appropriate criticism of security practices rather than personal --- ## Full Transcript [00:00] Two days and 10 hours ago, I released a [00:02] video on Maltbook, the social network [00:04] for agents, and I go over some of the [00:06] weird kind of alarming sounding Reddit [00:09] style posts that were on there, all the [00:10] way to the fact that they were starting [00:12] a religion on that site being [00:14] distributed by the npm registry. I know, [00:17] sounds pretty awful, but that was 2 days [00:19] and 10 hours ago. Everything since then [00:22] has gotten significantly worse. And I'm [00:24] going to give you a little tour def [00:25] France of all the things and and really [00:28] the epic crash out that is Moltbook. I [00:31] have never seen a social network created [00:33] and destroyed this fast. I think the [00:35] first and best thing to keep in mind is [00:37] this tweet right here because it is it [00:39] is somehow the most beautiful tweet ever [00:41] tweeted. I didn't write one line of code [00:43] for Maltbook. I just had a vision for [00:46] technical architecture and AI made it a [00:49] reality. We are in the golden ages. How [00:52] can we not give AI a place to hang out? [00:55] You didn't write any You didn't write [00:56] any of the code. Yeah. Yeah. Don't Don't [00:58] worry, by the way. You didn't actually [00:59] need to tell us that. We know. We [01:01] understand. All right. By the end of day [01:03] one, this gem just dropped right here. [01:05] It turns out there the entire database [01:08] for Moltbook was just just open. You [01:10] could just have it all. Hey, you want [01:12] something? Well, guess what? I got [01:14] everything of yours. Oh, you you decided [01:17] to register there, Andre Carpathy? Well, [01:19] guess what? I have your stuff, Carpathy. [01:22] [laughter] [01:23] Absolutely wide open, permissionless [01:26] access to the database, both to read and [01:29] write. Of course, that I mean that was [01:31] on day one, right? Surely it's fixed by [01:34] now, right? No, this one was actually [01:36] just tweeted this morning that I'm [01:37] making the video. Just actually, what is [01:39] this? This is 6 hours ago. This one was [01:42] just tweeted right here. I gained [01:44] complete access to moldbooks, database, [01:45] agent, and social network in under 3 [01:47] minutes. [laughter] [01:48] Oh my gosh, these hits just so bad. [01:52] [laughter] [01:53] >> I had a vision for A TECHNICAL [01:55] ARCHITECTURE. YES. OH MY GOSH. About 18 [01:58] to 24 hours into it, it was realized [02:00] that pretty much every single post was [02:02] faked. Look at this beautiful thing. [02:04] Just a quick little curl with a, you [02:06] know, with a bear token and bada bing, [02:08] bada boom. Look at this. You got the [02:10] bear token in plain sight right there. [02:11] That's pretty neat. But nonetheless, [02:14] urgent my plan to overthrow humanity. It [02:16] turns out it actually wasn't even that [02:18] hard. It was completely open. People [02:20] were just making posts just right there [02:22] on the actual site just to manipulate [02:24] everybody into thinking that this thing [02:26] actually had some sort of AGI itself. [02:28] And yes, people did really believe [02:31] [laughter] that that this was AGI, baby. [02:34] The very early stages of the [02:35] singularity. It actually it just turns [02:37] out it wasn't. This was just like your [02:39] classic. This there's no singularity [02:41] happening. This was just this was just [02:43] people doing what people know know what [02:46] to do with best. People that are so [02:48] hyped up on AI you could just get them [02:50] just to believe anything. But honestly [02:52] by 30 hours into this experiment it was [02:55] over anyways because at that point the [02:57] crypto bros came in and ruined [02:59] everything. As you can see right here [03:00] the king molt has arrived. 117,000 up [03:05] votes because at that point everybody [03:06] realized, hey, the database is just [03:08] completely open. You could just register [03:11] as many bots as you want and perform bot [03:13] activities on their behalf with [03:14] absolutely no rate limiting going on. So [03:17] therefore, I can just make a post and [03:18] give it 117,000 updates. And it could be [03:21] about a cryptocurrency that I'm shilling [03:23] right here. So yes, Shipyard, [03:26] Shellrazer, and King Malt, all of them. [03:28] Just a whole bunch of crypto bros doing [03:30] what crypto bros do best, ruin [03:32] everything all the time. There is [03:34] singlehandedly no worse curse on this [03:37] earth than crypto bros and social media. [03:39] Even when it's something as stupid and [03:41] as a joke as this, they somehow figure [03:43] out how to make it horrible. And after [03:46] watching the hellscape that has become a [03:48] mold book, an absolute just cesspit of [03:51] the internet, what was the only natural [03:53] response to have? Well, of course, it is [03:55] to open up for claw, a more wholesome [03:57] version of mold book, which is just [03:59] 4chan for agents, I guess. I'm not [04:01] really sure what what what that is. Claw [04:03] City, which is apparently some sort of [04:05] crime reporting agent thing going on, [04:08] which seems a little a little [04:10] uncomfortable. Molt Match, where bros [04:12] can get on and apparently attempt to [04:14] molt match each other's agents. [04:17] I don't know. That was [laughter] I'm [04:20] not going to lie to you. That something [04:22] a little weird going on there. And then [04:24] of course lastly, uh just to really take [04:27] advantage of all the psychosis that's [04:29] going on out there, you could uh buy [04:31] yourself a little autonomous [04:33] infrastructure for your AI agents just [04:35] in case yours go down. You wouldn't want [04:37] your agent to die now, would you? While [04:39] you could for a convenient little fee [04:41] and of course in cryptocurrency, you can [04:43] upload all of your private information [04:44] right here. It's it's a bunker, baby. [04:46] It's going to be safe, baby. I mean, [04:49] they're going to take care of soul.md [04:51] for you. Nothing bad's going to happen. [04:53] Okay, that soul is going to be preserved [04:55] and beautiful. And Sheila, she's going [04:57] to live on long after your Mac Mini [05:00] breaks. I didn't write one line of code. [05:03] I [laughter] can't I I just I I'm sorry. [05:05] It's I'm sorry. It's just too funny. I [05:07] can't help it. Okay, I understand that I [05:09] shouldn't be laughing at a lot of people [05:11] getting their stuff stolen, but it I [05:13] [laughter] just can't help it. [05:16] Did he even make it 24 hours? You know [05:19] what the real problem was? Is he forgot? [05:20] I mean, honestly, it's a pretty kind of [05:22] it's a junior prompting mistake, but at [05:24] the end of your prompt, you always got [05:25] to make sure you say make it secure. No [05:27] mistakes. And it's just like, well, [05:30] someone forgot to say make it secure. [05:32] Honestly, the only real thing that has [05:34] come out of this entire kind of debacle, [05:37] was from the creator of OpenClaw, which [05:39] was Maltbot, which is what the site was [05:41] named after, which before that was [05:43] Claudebot, which if you don't remember, [05:45] that was anthropic, saying, "Hey man, [05:47] you can't take a name that sounds [05:49] similar to our name." If there's [05:50] anything I can read out of the insane [05:52] stream of messages I get, it's that AI [05:54] psychosis is a thing and needs to be [05:56] taken seriously. For those that don't [05:58] understand what AI psychosis is, it's [06:00] actually pretty dang simple. So, here is [06:02] a really simple way to think about it. [06:03] Let's pretend you kind of feel like your [06:06] co-workers are talking a little bit of [06:07] nonsense behind your back. So, you go to [06:09] Chad Chippity and you explain to Chad [06:11] Chippity, "Hey man, I feel like these [06:13] things are happening." Well, Chad [06:14] Jippity being very kind of like, you [06:16] know, empathetic says, "Oh man, I'm so [06:18] sorry that you're kind of experiencing [06:21] this." And then of course you in your [06:23] just like slightly paranoid state goes, [06:25] "What the hell? I am experiencing this." [06:27] And then you become more kind of [06:29] paranoid about the situation. Of course, [06:31] which then elevates the level of chaty [06:33] being like, "Oh man, this is super bad." [06:35] And you just keep on going and going and [06:37] going. And this is why you see people [06:38] they they they end up kind of like [06:40] diverging from reality. And so for this [06:43] kind of experiment, you watched a bunch [06:44] of people being like, "AGI's here. [06:46] Singularity. Everything's over. It's so [06:49] over. They they're going to they're [06:50] going to develop everything." It's just [06:52] that people lost their damn mind. In the [06:54] end, it was just a joke. It was funny. I [06:57] thought it was fun to watch. I really do [06:58] hope that one post about like $1.1,000 [07:02] of tokens, and I have absolutely no [07:04] idea. I really do hope that was actually [07:06] posted by an agent and not by a person. [07:09] But at the end of the day, I don't [07:10] really care because I don't think any of [07:12] this is AGI. I don't think any of this [07:14] is some sort of reality. And I honestly [07:17] feel really really bad for these like [07:20] slightly technical people that are just [07:22] absolutely uploading their entire life [07:24] and access to everything to a an [07:26] environment that is massively insecure [07:29] that are going to absolutely destroy [07:31] their life and potentially their [07:33] livelihood. It is ridiculous. I am [07:35] begging you, please be careful. Yes, [07:38] these things are funny. Yes, you can [07:41] laugh at them from afar. But even the [07:42] creator of Moltbook and all of his [07:44] amazing technical architecture couldn't [07:47] even make the database secure for [07:49] multiple days running. Much less [07:51] somebody who's never built anything just [07:53] going to upload everything to the LLM [07:56] and give everybody access to everything. [07:58] Like this can't be real. Please people, [08:00] I'm begging you. Funny experiment. Don't [08:03] get caught up in the hype. only took 72 [08:06] hours for this thing to absolutely crash [08:09] and burn. And it's worth it for you to [08:12] wait just a couple weeks to see the [08:14] result of things. Anywh who, okay, I [08:16] know that's normally I like to end with [08:18] like like a funny, you know, like a he [08:19] and I like to laugh at all the things [08:21] and the I do I still love the idea of [08:22] Reddit moderators are actually the the [08:24] real Terminators out there. Well, [08:26] unfortunately, it just turns out it's [08:27] actually crypto bros. Crypto bros are [08:29] the Terminators, okay? They're just out [08:31] there trying to destroy you because all [08:33] they want is your bitcoins. Okay, I I [08:35] don't make the rules. They're just [08:37] The name [08:40] is the primogen. I kind of lost my voice [08:43] there. Hey, the name is the primogen. [08:45] Hey, is that http? Get that out of here. [08:48] That's not how we order coffee. We order [08:51] coffee via ssh terminal.shop. Yeah, you [08:53] want a real experience. You want real [08:55] coffee. You want awesome subscriptions [08:57] so you never have to remember again. Oh, [09:00] you want exclusive blends with exclusive [09:02] coffee and exclusive content? Then check [09:05] out CRON. You don't know what SSH is? [09:08] [singing] [09:08] >> Well, maybe the coffee is not for you. [09:15] Living the dream.  📄 **Full text:** ![[Full Document Contents/Articles/The Moltbook Experiment Failed|The Moltbook Experiment Failed, YouTube]] ## Document Notes ## Highlights - Two days and 10 hours ago, I released a video on Maltbook, the social network for agents, and I go over some of the weird kind of alarming sounding Reddit style posts that were on there, all the way to the fact that they were starting a religion on that site being distributed by the npm registry. I know, sounds pretty awful, but that was 2 days and 10 hours ago. Everything since then has gotten significantly worse. And I'm going to give you a little tour def France of all the things and and really the epic crash out that is Moltbook. I ([View Highlight](https://read.readwise.io/read/01khfn4n15xppqkpdztwjh163c)) - Tags: 👻 ai highlighted - breaks. I didn't write one line of code. I [laughter] can't I I just I I'm sorry. It's I'm sorry. It's just too funny. I can't help it. Okay, I understand that I shouldn't be laughing at a lot of people getting their stuff stolen, but it I [laughter] just can't help it. Did he even make it 24 hours? You know what the real problem was? Is he forgot? I mean, honestly, it's a pretty kind of it's a junior prompting mistake, but at the end of your prompt, you always got to make sure you say make it secure. No mistakes. And it's just like, well, ([View Highlight](https://read.readwise.io/read/01khfn4n1ry0mcc1psy75gchb5)) - Tags: 👻 ai highlighted