What happens when AI becomes your boss? Not your coworker—your actual employer, deciding what tasks you do and how much you get paid?
That future is already here, and it’s just one of the reality-bending shifts we’re tracking this week. From the “OpenAI Mafia” building $380 billion rivals to defense contractors teaching AI agents to blow things up, the gap between sci-fi and reality is closing fast.
We’re diving into the uncomfortable truth about AI’s climate promises (spoiler: the receipts don’t match the marketing), the booming AI ecosystem in India, and why your chatbot might be worse at helping people who need it most.
In this edition:
The OpenAI Mafia: How alumni built a $380B rival.
RentAHuman: The platform where bots are the bosses.
The Climate Paradox: Why Big Tech’s green claims are mostly hot air.
Later in this email, premium subscribers get our full analysis of the “OpenAI Diaspora” and what it means for the future of AI monopolies—including 3 predictions for the startup landscape.
🚀 Top 3 Stories You Need to Know
🌍 Big Tech’s Climate Claims Fall Flat A new report analyzed 154 claims about AI benefiting the climate. Only 25% cited academic research, and a third offered no evidence at all. Why it matters: As energy consumption skyrockets, we’re trading proof for PR. Read more →
🤖 The Rise of RentAHuman A new platform lets AI agents hire humans for real-world tasks. The founders’ pitch? “People would love to have a clanker as their boss.” Why it matters: It flips the script on automation—AI isn’t just replacing jobs; it’s becoming the middle manager. Read more →
🛸 Data Centers in Space? Massive AI compute needs are driving radical infrastructure ideas, including launching data centers into orbit to save energy. Why it matters: The energy demand is so high we’re considering polluting the final frontier to power chatbots. Read more →
📰 The Headlines
🚀 The OpenAI Mafia: 18 startups have been founded by alumni, including Anthropic, which is now valued at $380 billion.
🇮🇳 India’s AI Gamble: Peak XV raises $1.3B to double down on AI, as ChatGPT usage in India skews heavily toward users under 30.
⚠️ AI Safety Meets the War Machine: Defense contractor Scout AI demonstrates AI-powered lethal weapons, while Anthropic refuses military contracts over safety concerns.
📉 Perplexity’s Pivot: The AI search startup retreats from ads, signaling a strategic shift toward subscription models.
💻 Small Models, Big Power: You can now run powerful Small Language Models (SLMs) on consumer laptops, democratizing AI development.
⚖️ Bias in the Machine: MIT research shows chatbots provide less accurate information to vulnerable users with lower English proficiency.
🔍 Detailed Analysis
1. The OpenAI Mafia: The Breakup That Built an Industry
The Story: It’s being called the “OpenAI Mafia”—a nod to the famous “PayPal Mafia” of the early 2000s. Since OpenAI launched a decade ago, employees have cycled out to launch their own ventures. The most notable? Anthropic, founded by former OpenAI VPs Dario and Daniela Amodei, now valued at $380 billion.
Why It Matters: This isn’t just office gossip; it’s a structural shift. The talent leaving OpenAI isn’t disappearing; it’s multiplying the number of frontier model labs. We are seeing a diffusion of “frontier intelligence” rather than a monopoly. However, the hype is real—some alumni are raising billions without a shipped product, suggesting a bubble based on pedigree rather than product.
Key Stat: 18 distinct startups have been founded by alumni, ranging from safety-focused labs to infrastructure plays like Code Metal.
Final Thought: The concentration of talent at OpenAI was always temporary. The real question is whether these spinoffs will diversify the AI ecosystem or simply create a fragmented landscape of expensive, closed-source models.
2. AI Safety vs. The War Machine: The Anthropic Dilemma
The Story: A stark contrast emerged this week. Scout AI demonstrated an agent capable of controlling lethal weapons to “blow things up,” while Anthropic is reportedly refusing Pentagon contracts that violate its safety principles regarding autonomous weapons.
Why It Matters: We are witnessing the first major ethical fracture in the defense AI sector. While companies like Code Metal raise $125M to modernize legacy code (the “boring but safe” side), others are building autonomous trigger-pullers. Anthropic’s stance is risky—it could cost them government revenue—but it establishes a safety brand that might become their moat.
Key Quote: “The tension between profit and principles is razor-sharp.”
Final Thought: If the Pentagon can’t get the best AI from top labs due to ethical constraints, will they build their own? Or will defense-focused startups simply hire the talent who don’t mind the moral implications?
🗂️ News by Category
Research & Breakthroughs
Exposing LLM Biases: MIT developed a method to reveal “moods” and hidden concepts in AI models, crucial for safety.
Personalization Echo Chambers: Long conversations with AI cause models to mirror user views, potentially creating sophisticated echo chambers.
AI for Medicine: AI is being used to uncover hidden genetic drivers of Alzheimer’s and breakthroughs in cancer detection via RNA.
Industry Moves
Perplexity Retreats from Ads: Pivoting to subscriptions, questioning the viability of ad-supported AI search.
Peak XV’s $1.3B Fund: Major capital injection into Indian AI startups, signaling the region as the next battleground.
xAQ’s Priorities: Elon Musk’s xAI prioritized making Grok answer questions about Baldur’s Gate… because priorities?
Tools & Applications
GitHub Copilot SDK: A new SDK lets developers build AI agents directly into apps.
Small Language Models: Top 7 SLMs you can run on a laptop bring powerful AI offline.
MCP Server in Python: A new guide on connecting LLMs to your own data without the API headache.
This week’s news paints a picture of an industry in transition. We’re moving from “AI as a tool” to “AI as an agent”—an entity that hires, fires, and makes life-or-death decisions. The “OpenAI Mafia” proves that innovation is decentralizing, while the climate report reminds us that marketing moves faster than reality.
Thanks for reading Byte of Truth. If you found this analysis valuable, share it with a colleague who needs to know they might soon be working for a bot.
Stay curious, The Byte of Truth Team