Best of ChatGPT — October 2025
- 1
- 2
Hacker News·31w
“You have 18 months”
Explores the concern that AI's rapid advancement threatens not through job displacement, but through cognitive atrophy. As students increasingly use ChatGPT for writing and reading comprehension declines to 32-year lows, the real risk is humans deskilling themselves by outsourcing thinking to machines. Writing and reading form the foundation of deep symbolic thinking—the competitive advantage in knowledge work. Rather than worrying about AI surpassing human capabilities in 18 months, focus on maintaining the patience and discipline to engage in sustained, complex thought.
- 3
The Register·30w
ChatGPT: so popular, hardly anyone will pay for it
OpenAI is losing three times more money than it earns, with 95% of ChatGPT's 800 million users not paying for the service. The company posted a $13.5 billion net loss in the first half of 2025 against $4.3 billion in revenue, while committing over $1 trillion in datacenter spending through partnerships with Nvidia, AMD, and others. Despite ChatGPT generating 70% of OpenAI's recurring revenue and dominating 80% of generative AI web traffic, the company faces significant challenges converting free users to paid subscribers, with only 5% currently paying compared to an industry average of 3%.
- 4
Hacker News·28w
Stop Citing AI
Large language models like ChatGPT, Claude, and Gemini predict likely word sequences rather than provide factual information. These AI systems can generate convincing-sounding responses, but they lack source attribution and may produce inaccurate or unreliable information through hallucinations. Treating LLM outputs as authoritative sources is problematic, as they represent common word patterns rather than verified truths. The piece emphasizes the risks of over-trusting AI-generated content, particularly in critical domains like medicine and law.
- 5
rryyqn·29w
ChatGPT Atlas. The AI browser.
OpenAI launched ChatGPT Atlas, an AI-integrated web browser that enables instant AI assistance through text highlighting, remembers browsing context, and offers AI agent capabilities for paid users to automate tasks like booking trips and sending emails. The browser aims to make AI more accessible during web browsing with customizable privacy settings.
- 6
Fireship·31wOpenAI just made your entire tech stack obsolete...
OpenAI announced several new features at their dev day, including ChatGPT apps platform with 800 million weekly active users, Agent Kit for building AI workflows without extensive coding, GitHub Actions integration for automated code reviews, and API access to GPT-5 Pro and Sora 2. The updates include smaller, cost-effective models for voice and image generation, positioning ChatGPT as a potential operating system for app interactions.
- 7
Hacker News·31w
The AI bubble is 17 times the size of the dot-com frenzy - and four times subprime, this analyst argues
MacroStrategy Partnership argues AI represents a bubble 17 times larger than the dot-com era and 4 times bigger than the 2008 housing crisis, based on Wicksellian economic theory measuring capital misallocation from artificially low interest rates. The analysis claims large language models have hit scaling limits, citing ChatGPT-5's $5 billion cost with minimal improvement over ChatGPT-4, low task completion rates at companies (1.5-34%), and declining AI adoption among large enterprises. The firm predicts this will trigger a deflationary recession similar to the early 1990s S&L crisis, recommending investors shift away from AI companies toward resources, emerging markets, gold, and short-dated Treasuries.
- 8
Fireship·29wOpenAI’s new browser feels familiar…
OpenAI released Atlas, an AI-powered browser built on Chromium that integrates ChatGPT as an assistant capable of viewing browsing history and performing actions on behalf of users. The browser faces similar security challenges as other AI browsers, particularly vulnerability to prompt injection attacks. While Atlas offers convenience through agent mode for tasks like ordering food, it raises privacy concerns despite user controls for managing browsing memory.
- 9
Collections·30wOpenAI to Introduce Adults-Only Features for ChatGPT by December
OpenAI will allow ChatGPT to support adult content for verified users starting December 2024, reversing its previous content restrictions. The company established an Expert Council on Well-Being and AI to address mental health, age verification, and safety concerns. New features include age-gating measures, customizable chatbot personalities, and tools to detect mental distress while maintaining user safety protections.
- 10
Medium·28w
Designing the Jarvis moment
OpenAI's Apps SDK enables third-party applications to integrate seamlessly within ChatGPT conversations, creating contextual experiences where users complete tasks without switching interfaces. The SDK applies Hick's Law and Fitts's Law principles to reduce decision complexity and interaction distance. Designers should focus on single-purpose, conversation-friendly tasks that can be summarized visually with minimal actions. Best practices include displaying only relevant information, limiting cards to two primary actions, and avoiding complex multi-step workflows. This shift positions ChatGPT as an operating system-like environment, expanding UX design scope toward flows, contexts, and systems that help AI communicate and align with human goals.
- 11
Where's Your Ed At·32w
OpenAI Is Just Another Boring, Desperate AI Startup
Critical analysis of OpenAI's business strategy, arguing the company lacks focus and direction despite massive funding. The piece examines OpenAI's scattered product announcements across social media, productivity tools, hardware, and advertising, while highlighting that ChatGPT subscriptions remain its primary revenue source. The author contends OpenAI operates like a typical AI startup with unsustainable R&D spending, commoditized products, and inherent technical limitations like hallucinations. Revenue growth is reportedly slowing while costs exceed income, with the company spending 150% of H1 2025 revenue on R&D that produced underwhelming results like GPT-5 and expensive-to-operate Sora 2.
- 12
Product Hunt·30w
nanochat: The best ChatGPT that $100 can buy
nanochat is a minimal, full-stack LLM implementation by Andrej Karpathy in approximately 1000 lines of code. It enables running the complete pipeline—tokenization, pretraining, finetuning, evaluation, inference, and web UI—on a single 8XH100 node for under $1000. The project achieves competitive performance at the $100 tier model level while maintaining clean, hackable code designed to make LLM development accessible for learning purposes.
- 13
Get to the Top·28w
AI Isn’t Replacing Writers. It’s Replacing the Internet.
Oxford data reveals that over 50% of online text is now AI-generated, up from 5% in 2020, with projections suggesting 90% by next year. This shift raises concerns about model collapse—where AI systems trained on AI-generated content produce increasingly degraded output—and the "dead internet theory," where automated systems dominate online discourse. The trend prioritizes cost and speed over originality and truth, potentially creating a feedback loop of recycled, low-quality content that may ultimately drive demand for authentic human voices.
- 14
Collections·29wOpenAI Launches ChatGPT Atlas: A New AI-Powered Browser
OpenAI launched ChatGPT Atlas, an AI-powered web browser for macOS with plans for cross-platform support. The browser features autonomous task execution, real-time page analysis, and contextual memory capabilities. Security researchers have identified vulnerabilities including CSRF and prompt injection attacks that could enable malicious command execution and data exfiltration. Atlas enters a competitive market alongside Google Chrome and emerging AI browsers like Perplexity's Comet, positioning OpenAI in the search advertising space while facing ongoing security and adoption challenges.
- 15
UX Planet·31w
UI Design with ChatGPT 5
Demonstrates how to use ChatGPT 5 to generate UI designs for a mobile food delivery app through structured prompts. Covers the complete workflow from writing detailed markdown prompts to generating visual mockups, exporting to Figma, and producing working code. Includes practical tips for effective prompt writing, such as using markdown formatting, describing layout and style explicitly, and maintaining clear hierarchy. Shows improvements over ChatGPT 4o in following design specifications and producing cleaner outputs.
- 16
The Register·29w
AI eats leisure time, makes employees work more, study finds
Research analyzing US Bureau of Labor Statistics data from 2004-2023 reveals that workers in occupations with higher AI exposure work an average of 3.15 additional hours per week while experiencing 3.20 fewer leisure hours. Despite productivity gains from AI tools like ChatGPT, the economic benefits primarily flow to employers and consumers rather than workers, who report lower job satisfaction and longer hours due to increased surveillance and productivity demands. The study challenges the assumption that AI automation reduces workload, showing instead that competitive labor markets prevent workers from capturing technology-driven productivity gains.