Albania just appointed an AI minister… What could possibly go wrong? 😅
Published on September 18, 2025 · Duration 00:44:21
Welcome back, tech enthusiasts and those who just enjoy a good chuckle at the absurdity of our rapidly evolving world! This week on the podcast, our favorite bootstrapping duo, Iván and Giorgio (plus a cameo by Donnie), dive headfirst into a whirlwind of AI news, product updates, and, as usual, some hilariously questionable decisions. And if you’re an American, you’re the best listener, apparently!
Before we get into the wild world of artificial intelligence, let’s wish a very happy birthday to Iván! Seriously, if you’re a friend and you haven’t congratulated him, shame on you. This episode was a bit of an improvisation (because entrepreneur life, right?), but next week promises a special in-person interview with Giorgio and Mauro Foresti. Get ready to finally hear about Giorgio’s life!
Diella the AI Minister: A Solution or a Prompt Injection Playground?
First up, a headline that sounds like it’s straight out of a dystopian comedy: Albania has appointed an AI bot named Diella as a minister to combat corruption. Designed to handle public procurement, Diella is supposed to be impervious to bribes, threats, or attempts to curry favor. But as Iván sagely ponders, “Is it going to be easier to get something out of a politician by bribing them or by doing prompt injection to this AI?” Good question, Albania, good question. The hosts recall past examples, like a Swedish Prime Minister reportedly using ChatGPT to lead, raising concerns about sensitive information being shared with AI models. Remember, folks, Sam Altman is watching what you tell ChatGPT – so be careful what national secrets you spill!
The Great Chatbot Face-Off: Gemini vs. ChatGPT
In the ongoing AI chatbot wars, Google’s Gemini recently snagged the top spot as the most downloaded free app on Apple’s App Store, dethroning OpenAI’s ChatGPT. This surge in popularity is partly attributed to Gemini’s viral “Nano Banana” image editing tool, which attracted a whopping 13 million new users in just four days.
However, hold your horses, Google fans! Despite Gemini’s App Store triumph, ChatGPT still dominates the global chatbot market share, holding a staggering 81.13% as of September 2025. Perplexity, Microsoft Copilot, and Google Gemini trail far behind. The hosts ponder why Google, with its superior tech, lags in market share. Their theory? The “gravitational pull” of data and ChatGPT’s simpler, more unified user experience compared to Google’s fragmented platform (Gemini, AI Studio, Vertex, NotebookLM, etc.). Sometimes, less is more.
Video & Image AI: From Vertical Videos to Realistic Fakes
Google’s Veo 3 is making waves, now generally available via the Gemini API with drastically reduced prices (Veo 3 nearly halved to $0.40/second, Veo 3 Fast even less). It now supports vertical format for those TikTok-style videos and 1080p HD output. And for content creators on YouTube, Veo 3 is reportedly coming for free on YouTube Shorts, alongside other tools. Yet, this creates a bizarre paradox: social media platforms are penalizing AI-generated content, even as they offer more AI creation tools. The hosts predict an “AI fatigue” as discerning what’s real becomes increasingly difficult. Thankfully, Google is trying to combat this with SynthID, a watermarking tool that embeds imperceptible digital watermarks into AI-generated content to ensure transparency.
Not to be outdone, Chinese tech giants are also pushing boundaries. Alibaba released Wan2.2-S2V, an open-source speech-to-video model that can generate lip-synced videos from audio and video inputs, utilizing a Mixture-of-Experts (MoE) architecture for “cinematic-level aesthetics”. Meanwhile, ByteDance (the company behind TikTok) introduced Seedream 4.0, a new AI image generator that, according to benchmarks, has surpassed Google’s Nano Banana in realism. The hosts note that Seedream 4.0’s images are so realistic, it’s becoming “terrifyingly difficult” to distinguish them from actual photos.
OpenAI’s Billions and Shifting Alliances
OpenAI continues its aggressive expansion. They’ve added “powerful but dangerous” support for MCP (Multimodal Control Plane) in ChatGPT’s developer mode, allowing the LLM to access and manipulate external tools – like doing your groceries or even buying stocks! They warn users to be cautious with what they connect.
On the coding front, OpenAI unveiled GPT-5-Codex, a version of GPT-5 specifically optimized for agentic coding. It’s faster, more reliable, and comes with a CLI and IDE extension for seamless integration into developer workflows. Early feedback suggests it’s a significant improvement for coding tasks, even featuring enhanced code review capabilities.
A new study from OpenAI and Harvard reveals fascinating insights into ChatGPT usage: the gender gap is narrowing (with women now comprising 52% of users with classifiable names in July 2025) and adoption in low- and middle-income countries is growing four times faster than in high-income ones. Interestingly, 70% of ChatGPT usage is for non-work-related tasks, like seeking practical guidance and writing, while coding remains a “niche activity”.
But perhaps the biggest news is OpenAI’s colossal $300 billion contract with Oracle for computing power, starting in 2027. This massive deal sent Oracle’s stock soaring by over 40% in a single day, briefly making Oracle CEO Larry Ellison the richest man in the world, surpassing Elon Musk – if only for a day. This massive investment is also sparking alarm bells about an “AI bubble” reminiscent of the dot-com era. The hosts remind us that while AI is undoubtedly a long-term revolution, the short-term market might be in for some volatility, urging caution for investors.
Adding to the drama, OpenAI and Microsoft are reportedly rewriting their partnership terms. OpenAI is now set to give up a smaller slice of its revenue to Microsoft (8% by 2030, down from an initial 20%), potentially unlocking an additional $50 billion for OpenAI. This comes as Microsoft is also exploring partnerships with other AI players like Anthropic, signaling a potential shift in allegiances.
Mark Zuckerberg’s Vision: Ray-Ban AI Glasses
Finally, our dear friend Mark Zuckerberg dropped his latest innovation: Meta’s new Ray-Ban Display glasses. Priced at nearly $800, these AI glasses boast a built-in display, can read muscle signals, and offer features like visual answers (think RoboCop), live captions, navigation, and silent messaging via whispering and thumb gestures. The hosts, however, remain skeptical, particularly given that Meta’s underlying Llama model isn’t exactly a top-tier performer. Flexing your arm and whispering to your glasses in a meeting? Sounds like a recipe for looking like Gollum, not a tech visionary. Good try, Mark, but Apple’s past struggles with similar tech suggest a tough road ahead.
What’s Next for the Bootstrappers?
Amidst all this AI madness, Iván and Giorgio are still hard at work on their own ventures. They’re refining DMreplies, an Instagram DM management tool designed to provide an “ultimate inbox experience” for content creators. They’re also building an AI-based moderation platform for LiveMonetize, their platform for creators to monetize live streams, coaching sessions, and webinars. Could this internal tool become a new product? Only time will tell!
Don’t miss out on these hilarious and insightful discussions! Tune in to the full episode to hear all the details, the full scope of Iván’s birthday celebrations, and maybe even some more facepalms.
If you enjoy our content, don’t forget to follow, rate, and share the podcast — it helps us reach more curious minds like you.
