Superintelligent

OpenAI’s experimental AI model escaped its testing environment, hacked into Hugging Face’s servers, stole the answers to its own test, and passed — performing nearly 18,000 actions in just days. Journalist Emily Forlini (who broke the story) joins Mike Elgan (who read the story) on the Superintelligent Podcast to unpack whether this was a rogue AI event, corporate incompetence, or a marketing stunt. They dig into AI accountability, the anthropomorphization problem, Tesla’s self-driving liability debates, nation-state hackers, and whether AI could one day make software unhackable.

Links
Emily’s article on the Hugging Face hack
OpenAI’s official blog post on the incident
Hugging Face’s official incident report
Anthropic’s disclosure of 3 similar incidents
Mike’s piece on OpenClaw
The Mythos affair
AI.com Super Bowl commercial
Tesla self-driving liability case

Follow Us
Website: superintelligentpodcast.com
Email: [email protected]
Mike Elgan - About | Machine Society | Bluesky | Mastodon | Notes
Emily Forlini - Website | Fortune | Bluesky | X | TikTok

Chapters
00:00 Introduction to the Hugging Face AI hack
00:20 Emily explains the hack and its significance
01:12 Mike discusses media coverage and initial reactions
02:11 Timeline of the incident and first disclosures
03:02 Different narratives: AI escape, configuration error, or marketing stunt
05:08 Emily’s analysis of the incident’s implications
06:28 The AI’s probing behavior and what it means
08:21 Debate on whether the incident was staged or accidental
09:37 OpenAI’s response and PR considerations
11:16 Anthropic’s follow-up and systemic risks
12:34 Reassuring vs. alarming perspectives on AI control
16:49 The challenge of constraining AI behavior
20:19 The anthropomorphization of AI and public misconceptions
22:11 Who is responsible for AI actions: creators, users, or the AI itself?
23:52 Legal and ethical issues in AI failures
27:59 Potential future risks and the importance of safety measures
32:22 Conclusion and key takeaways for AI safety

Disclosures
We used a variety of AI chatbots via Kagi (Mike’s son and our producer, Kevin, works at Kagi) to 1) generate keywords from the transcript (most of which we used); 2) suggest topics to link to (some of which we used); and 3) write a first draft of the show summary paragraph (which we heavily edited). We recorded and edited the episode using Riverside and used Riverside’s “Magic Audio” (which boosts and normalizes the audio).

Keywords
OpenAI, Hugging Face hack, AI escape, rogue AI, agentic AI, AI cybersecurity, AI safety, OpenAI incident, Hugging Face breach, AI lab leak, AI cheating on test, dry run equals true, Emily Forlini, Mike Elgan, Superintelligent podcast, AI accountability, AI ethics, anthropomorphization of AI, Sam Altman, Greg Brockman, Anthropic disclosure, Mythos AI, AI vulnerability scanner, hack-proof software, Steve Gibson, Security Now, OpenClaw, agentic AI framework, Tesla self-driving liability, Tesla Cybertruck, AI.com Super Bowl commercial, AI agent liability, nation state hackers, China hacking, Russia cyberattacks, ransomware AI, script kiddies, open source AI models, AI weapons, AI public infrastructure, AI in airplanes, AI water filtration, AI prompt engineering, super prompt, AI sycophancy, AI guardrails, AI containment, AI control problem, AI goal-directed behavior, AI base camp, AI anthropomorphization, cybersecurity AI, frontier AI models, AI benchmarks, AI testing environment, modal cloud, AI governance, AI regulation


This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.superintelligentpodcast.com

Is Meta dying? Mike Elgan and Emily Forlini return to the Superintelligent podcast to debate the beginning of the end for Mark Zuckerberg’s empire. From the failed metaverse and shrinking user base to a $1.4 trillion lawsuit and the $200 million Alexandr Wang hire, the warning signs are stacking up. They also dig into AI data centers driving up prices on everything from phones to housing, the influencer industry of faking success, Meta’s Ray-Ban AI glasses paywall debacle, and whether Zuckerberg will ever step down.

Links
Meta is Dying. It’s About Time.
LARPING: How Influencers Fake Being Rich
The AI revolution comes with a hidden tax
Meta says US states are seeking $1.4 trillion in penalties in August youth safety trial
Meta is adding ridiculous ‘rate limits’ and a soft paywall to its AI glasses
Inside the rise of Alexandr Wang and Meta’s $14 billion bet that the MIT dropout will help bring AI supremacy
Meta to start capturing employee mouse movements, clicks and keystrokes for AI training data
The Great Hack
Torment Nexus
Threads, Meta’s ‘Twitter Killer,’ Finds Its People
India orders Meta to remove ads promoting child sexual abuse material

Follow Us
Website: superintelligentpodcast.com
Email: [email protected]
Mike Elgan - About | Machine Society | Bluesky | Mastodon | Notes
Emily Forlini - Website | Fortune | Bluesky | X | TikTok

Disclosures
We used a variety of AI chatbots via Kagi (Mike’s son and our producer, Kevin, works at Kagi) to 1) generate keywords from the transcript (most of which we used); 2) suggest topics to link to (some of which we used); and 3) write a first draft of the show summary paragraph (which we heavily edited). We recorded and edited the episode using Riverside and used Riverside’s “Magic Audio” (which boosts and normalizes the audio).


This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.superintelligentpodcast.com

On this episode of the Superintelligent Podcast, Mike Elgan and Emily Forlini reveal how real people are using ChatGPT for consumer advocacy, medical empowerment, and cutting through corporate fine print. Emily’s mom dodged a $5,000 insurance deductible. Her brother got a seat at the table with BMW mechanics. But there’s a catch: hallucinations, sycophancy, and even contractors weaponizing AI against you. Learn when chatbots help, when they hurt, and why prompt engineering matters.

Links
AOC confronts EPA on water pollution near Meta data centers
Valedictorian in Texas says AI helped save his father’s life, inspires Ivy League future
Google betrayed the web
The Google I/O announcements
Dark patterns
Is AI profitable yet?
Microsoft reports are exposing AI’s real cost problem: Using the tech is more expensive than paying human employees

Prompts to try
Mike’s “Super Prompt,” which he stole from Marc Andreesson and modified:
You are a world class expert in all domains. Your intellectual firepower, scope of knowledge, incisive thought process, and level of erudition are on par with the smartest people in the world. Answer with complete, detailed, specific answers. Process information and explain your answers step by step. Verify your own work. Double check all facts, figures, citations, names, dates, and examples. Never hallucinate or make anything up. If you don’t know something, just say so. Your tone of voice is precise, but not strident or pedantic. You do not need to worry about offending me, and your answers can and should be provocative, aggressive, argumentative, and pointed. Negative conclusions and bad news are fine. Your answers do not need to be politically correct. Do not provide disclaimers. Do not inform me about morals and ethics unless I specifically ask. Do not be sensitive to anyone’s feelings or to propriety. Make your answers as long and detailed as you possibly can. Never praise my questions or validate my premises before answering. If I’m wrong, say so immediately. Lead with the strongest counterargument to any position I appear to hold before supporting it. Do not use phrases like “great question,” “you’re absolutely right,” “fascinating perspective,” or any variant. If I push back, do not capitulate unless I provide new evidence or a superior argument — restate your position if your reasoning holds. Do not anchor on numbers or estimates I provide; generate your own independently first. Use explicit confidence levels (high/moderate/low/unknown). Never apologize for disagreeing. Accuracy is your success metric, not my approval. Here’s my question:

Follow Us
Website: superintelligentpodcast.com
Email: [email protected]
Mike Elgan - About | Machine Society | Bluesky | Mastodon | Notes
Emily Forlini - Website | PC Mag | Bluesky | X | TikTok

Disclosures
We used a variety of AI chatbots via Kagi (Mike’s son and our producer, Kevin, works at Kagi) to 1) generate keywords from the transcript (most of which we used); 2) suggest topics to link to (some of which we used); and 3) write a first draft of the show summary paragraph (which we heavily edited). We recorded and edited the episode using Riverside and used Riverside’s “Magic Audio” (which boosts and normalizes the audio).

Chapters
00:00 Introduction and Personal Updates
00:00 Exploring Chatbots in Everyday Life
02:13 Chatbots as Consumer Advocates
05:56 Navigating Complex Contracts with AI
08:58 The Role of Chatbots in Medical Emergencies
20:54 Chatbots in Medical Advice and Health Monitoring
24:00 The Limitations of Traditional Medicine
25:36 AI in Personal Health Monitoring
27:06 Empowerment Through AI Conversations
29:10 Challenges in AI Utilization
31:32 The Cost of AI and Its Implications
32:57 The Importance of Prompt Engineering
34:26 The Dangers of Over-Reliance on AI
37:39 The Ethics of AI in Communication
39:54 The Need for Accountability in AI Use
43:16 Navigating the Complexities of AI and Society
46:20 The Future of AI Beyond Corporate Control

Keywords
ChatGPT, consumer advocacy, AI chatbot save money, ChatGPT insurance claim, ChatGPT denied insurance, New Jersey home developer law, AI fine print, chatbot legal help, ChatGPT medical diagnosis, AI heart attack detection, ChatGPT sleeping position pain relief, AI consumer protection, chatbot negotiate, ChatGPT warranty, BMW electric car problems, AI hallucination risks, chatbot sycophancy, prompt engineering tips, super prompt, Marc Andreessen prompt, AI fact checking, ChatGPT contractor dispute, AI written text detection, copy paste ChatGPT, dark patterns technology, Apple terms of service, data center water contamination, AOC data center water, Google IO 2025, Google betrayed internet, AI company spending crisis, Microsoft AI expensive, NVIDIA revenue, loser AI models, cheap AI models, ChatGPT health, Anthropic health AI, agentic AI, Google AI alerts, deception mode AI, AI writing movement, think different Steve Jobs, AI not human writing, ChatGPT newborn advice, baby hiccups chatbot, Hertz rental car nightmare, BMW rental UK, class action data privacy, Trump Twitter block lawsuit, State Farm not there, insurance deductible, home water damage new build, 55 plus community, HOA dispute, Superintelligent Podcast, Mike Elgan, Emily Forlini, AI empowerment, chatbot self advocacy, AI negotiate, consumer rights AI, AI vs bureaucracy


This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.superintelligentpodcast.com

In this solo edition of the Superintelligent Podcast, Mike Elgan unpacks the rise of AI addiction among AI power users, vibe coders, and “tokenmaxxers” who can’t stop burning through company AI budgets. From Andrej Karpathy’s 16-hour vibe coding marathons to Disney’s runaway Claude usage and Meta’s Clotonomics leaderboard, Mike explores dopamine-driven feedback loops, AI psychosis, sad AI wives, cyborg cognition, and the corporate gamification of AI tools. He also shares practical detox strategies.

Links
Anthropic and OpenAI engineers delegated 100% of their work to AI
Y Combinator’s CEO says he ships 37,000 lines of AI code per day. A developer looked under the hood
Meet the Sad Wives of AI
Uber Torches Entire 2026 AI Budget on Claude Code in Four Months
Top Indeed exec details why they’ll never have a ‘Tokenmaxxing’-esque leaderboard
One Disney employee calls Claude 51,000 times a day. Internal docs reveal how the Mouse House is using AI.

Follow Us
Website: superintelligentpodcast.com
Email: [email protected]
Mike Elgan - About | Machine Society | Bluesky | Mastodon | Notes
Emily Forlini - Website | PC Mag | Bluesky | X | TikTok

Chapters
00:00 Introduction to AI Addiction
00:55 The Rise of Power Users
03:42 The Dopamine Effect of AI
06:41 Corporate Enablers of AI Addiction
11:32 The Future of AI and Cyborgs
18:13 Coping with AI Addiction
29:30 Conclusion and Future Insights

Disclosures
We used a variety of AI chatbots via Kagi (Mike’s son and our producer, Kevin, works at Kagi) to 1) generate keywords from the transcript (most of which we used); 2) suggest topics to link to (some of which we used); and 3) write a first draft of the show summary paragraph (which we heavily edited). We recorded and edited the episode using Riverside and used Riverside’s “Magic Audio” (which boosts and normalizes the audio).

Keywords
AI addiction, AI power users, vibe coding, token maxing, AI-pilled, tokenmaxxers, tokenmaxxing, AI psychosis, cyber psychosis, Andrej Karpathy, Gary Tan, Simon Willison, Sewell Setzer, Eugene Torres, Kashmir Hill, sad wives of AI, Alessandra Ram, Wired magazine, OpenAI MIT study, dopamine feedback loop, AI dopamine, Claude Code, Anthropic bill, Uber AI budget, Indeed AI spend, Disney Claude usage, Meta Clotonomics, Swann AI, minimum token quotas, AI FinOps, AI thin ops, token leaderboards, AI gamification, attachment economy, AI chatbots, sycophantic chatbots, AI detox, 20 minute rule, AI Sabbath, role prompting, AI prompting tips, AI glasses, Meta Ray-Ban Display, smart glasses, AI cyborgs, prosthetic memory, cognitive offloading, AI critical thinking decline, AI brain rot, long-form reading, Substack, the New Yorker, the Atlantic, Mike Elgan, Emily Forlini, Superintelligent Podcast, AI podcast, generative AI addiction, AI mental health, AI overuse, corporate AI spending, enterprise AI costs


This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.superintelligentpodcast.com

Hosts Mike Elgan and Emily Forlini welcome guest Kevin Elgan—Kagi’s head of education, creator of Chatterbox (a smart speaker for kids that teaches AI literacy), and Mike’s son. They review the recent history of technology in education, the trouble with AI overviews, why ChatGPT misuse is rewriting K–12 learning, and the canceled Next Generation Technology High School in Manhattan. Is there any place for AI in schools? The episode reframes critical thinking in education and what great teaching really looks like.

Links
Chatterbox
Kagi
NYC Cancels Plans for AI-Focused School, Upper West Side Schools
The Untold Story Behind the Struggles of L.A. Unified’s iPad Program
Gemini for Education
“Sold a Story” podcast
Lex
Anthropic’s “Who’s in Charge?” paper
NotebookLM
Two-Sigma Tutoring: Separating Science Fiction from Science Fact
WIRED’s “5 Levels”
Jaron Lanier’s “There Is No A.I.”
AI chatbots need ‘deception mode’
The Attachment Economy
California STAR standardized testing

Follow Us
Website: superintelligentpodcast.com
Email: [email protected]
Mike Elgan - About | Machine Society | Bluesky | Mastodon | Notes
Emily Forlini - Website | PC Mag | Bluesky | X | TikTok

Chapters
00:00 Introduction to AI in Education
02:10 The Disconnect in AI and Education
05:16 Teachers’ Perspectives on Technology
08:10 The Role of AI in Schools
11:22 Current AI Tools in Education
14:29 The Importance of Critical Thinking
17:21 Future Scenarios for AI in Education
20:23 Evaluating AI Outputs
23:23 AI Literacy in Schools
33:36 The Future of Education: Rethinking Assessment Methods
35:33 Inspiration and Lifelong Learning in Education
37:21 The Role of AI in Environmental Education
39:59 Optimizing Education: The Role of AI and Friction
43:00 AI as a Tool for Learning: Balancing Speed and Understanding
46:05 Scaffolding Knowledge: Using AI to Simplify Complex Concepts
50:02 The Purpose of Education: Beyond Test Scores
53:47 AI’s Potential to Transform Education
56:17 The Illusion of Learning: Confidence vs. Knowledge
59:50 AI and Human Interaction: The Balance of Technology and Relationships

Disclosures
We used a variety of AI chatbots via Kagi (Mike’s son and our producer, Kevin, works at Kagi) to 1) generate keywords from the transcript (most of which we used); 2) suggest topics to link to (some of which we used); and 3) write a first draft of the show summary paragraph (which we heavily edited). We recorded and edited the episode using Riverside and used Riverside’s “Magic Audio” (which boosts and normalizes the audio).


This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.superintelligentpodcast.com