AI·41wBREAKING: GitHub accidentally leaked GPT-5 details (proof inside)
GitHub accidentally published and quickly deleted a changelog entry announcing GPT-5's general availability in GitHub Models. An archived version of the deleted page serves as evidence of the premature announcement, suggesting GPT-5 may be launching imminently.
Where's Your Ed At·39wAI Bubble 2027
An MIT study reveals 95% of organizations get zero return from generative AI, while Meta freezes AI hiring and major outlets question if we're in a bubble. The analysis predicts the AI bubble will burst through a series of events over 18 months, including NVIDIA's growth slowing, AI funding drying up, major AI companies collapsing, and Big Tech pulling back from AI investments. Key vulnerabilities include OpenAI and Anthropic burning billions annually, CoreWeave's financial troubles, and AI startups raising at unsustainable valuations. The bubble is driven by vibes rather than returns, making it vulnerable to emotional market reactions when reality sets in.
Hacker News·41wgpt-5 leaked system prompt
A leaked system prompt reveals GPT-5's internal instructions and capabilities. The prompt shows personality guidelines emphasizing clarity and enthusiasm, memory management through a 'bio' tool, canvas functionality for document creation, image generation capabilities, Python code execution environment, and web search tools. It includes specific behavioral constraints like avoiding opt-in questions and copyright material reproduction.
Docker·41wGPT-5 Broke AI Apps: What Devs Must Do Now
GPT-5's launch caused widespread AI application failures when OpenAI deprecated older APIs without warning. The incident highlights the brittleness of AI systems that depend on single providers or models. Modern AI applications are complex stacks involving prompts, embeddings, and retrieval logic that break when underlying models change. To build resilient AI systems, developers should implement AI High Availability (AIHA) architectures with multi-provider support, automated fallback mechanisms, behavioral monitoring, and contract testing. Key strategies include abstracting API layers, maintaining separate prompt libraries for different models, implementing graceful degradation, and treating model deprecation as an expected lifecycle event rather than an emergency.
Fireship·41wGPT-5 is here... Can it win back programmers?
OpenAI released GPT-5, claiming it's the first AI to outperform humans on certain benchmarks, but the reality is more nuanced. While GPT-5 unifies multiple models for better task routing and costs significantly less than competitors at $10 per million tokens, it still has limitations in coding tasks. Testing shows it can generate functional Svelte applications but makes errors with framework-specific rules. The model represents more of a consolidation effort than a revolutionary breakthrough, and programmers' jobs remain safe for now.
Where's Your Ed At·42wHow Much Money Do OpenAI And Anthropic Actually Make?
OpenAI and Anthropic use annualized recurring revenue (ARR) metrics to inflate their apparent success, with OpenAI claiming $12 billion ARR and Anthropic $4-5 billion ARR. However, analysis of actual monthly revenues suggests OpenAI made around $3.6 billion in 2024 and $5.3 billion through July 2025, while Anthropic made approximately $1.5 billion through July 2025. ARR multiplies one month's revenue by 12, creating misleading impressions of annual performance and allowing startups to appear more successful than actual cash flow indicates.
The Register·40wAre you willing to pay $100k a year per developer on AI?
AI companies are currently selling their services at loss leader prices, but this won't last forever. While tools like GitHub Copilot claim to write 30% of code, many developers don't trust AI results and spend significant time fixing AI-generated errors. Current AI pricing of $20-200 per month per developer could increase 10-15 times when companies like OpenAI (burning $8 billion annually) need to become profitable. The real implementation costs often run double or triple initial estimates, making AI adoption far more expensive than anticipated.
Stefan Mischook·39wDevelopers Rejoice! The Ai Bubble is Bursting!
The AI hype cycle is normalizing as companies struggle to achieve expected ROI from AI investments. The job market for developers, particularly juniors, is returning to pre-COVID levels after an artificial boom period. While AI tools remain valuable for development, they require proper prompt engineering and edge case management to be effective. Recent changes to GPT models highlight the brittleness of AI development and the importance of maintaining backward compatibility in software systems.
Getting started with AI·40wAutoGen and MCP
Learn how to enhance AutoGen agents by connecting them to MCP (Model Context Protocol) servers, giving them access to external tools and capabilities. The tutorial demonstrates setting up a Python application with three agents that can communicate with both local and remote MCP servers, including a web fetching server and a custom C# server. Using the autogen-ext[mcp] extension, developers can easily integrate any MCP-compliant server to expand their agents' functionality beyond basic chat interactions.
Fireship·42wGoogle’s Genie model makes realistic worlds in realtime…
Google DeepMind released Genie 3, a world model that generates controllable virtual environments from text prompts in real-time at 720p/24fps. The model creates interactive worlds with physical properties for robot training simulations. OpenAI released GPT-O OSS under Apache 2.0 license, offering open-source reasoning capabilities that can run locally. Anthropic upgraded Claude Opus 4.1 with improved software engineering capabilities, particularly for multifile code refactoring in large projects.
habr·38wOpenAI's Codex CLI Agent: The Complete VS Code Setup Guide
A comprehensive guide for setting up OpenAI's Codex CLI agent extension in Visual Studio Code. Covers installation, authentication with ChatGPT Plus account, interface configuration including environment and mode settings, and demonstrates creating a complete to-do application with database and Bootstrap styling. Includes tips for data privacy settings and model selection.
Medium·38wGPT-5 System Prompt Leaked : 7 Prompt Engineering Tricks to learn
Analysis of a leaked GPT-5 system prompt reveals seven key prompt engineering techniques including identity locking to prevent prompt injection, knowledge anchoring for temporal context, multimodal toggles for routing, personality injection for behavioral control, content safety as first-class instructions, self-denial of hidden mechanisms to prevent conspiracy theories, and dynamic retrieval gates for up-to-date information. The techniques demonstrate advanced strategies for building robust AI systems through careful prompt design rather than fine-tuning.
Ars Technica·38wZuckerberg’s AI hires disrupt Meta with swift exits and threats to leave
Meta faces internal disruption as newly hired AI executives, including OpenAI's ChatGPT co-creator Shengjia Zhao, threaten to quit shortly after joining. Zuckerberg is conducting Meta's biggest leadership reorganization in 20 years, bringing in high-profile AI talent like former Scale AI CEO Alexandr Wang and former GitHub chief Nat Friedman. However, several new AI hires have already left after brief tenures, highlighting the challenges of integrating external talent into Meta's culture while competing in the AI race.
The Register·40wSam Altman admits that AI is a bubble, but still a big thing
OpenAI CEO Sam Altman acknowledged that AI is currently in a bubble phase, comparing it to the dot-com era where overexcitement led to inflated valuations despite underlying technological importance. He believes AI will survive the eventual burst, similar to how the internet persisted after the dot-com crash. Despite recognizing the bubble, Altman plans massive expansion, stating OpenAI will spend trillions on datacenter construction. The company faces GPU shortages that influenced ChatGPT-5's design focus on cost optimization rather than power. OpenAI's revenue reached $10 billion annually but the company still operates at a loss, raising questions about funding sources for ambitious expansion plans.
Ars Technica·39wIs the AI bubble about to pop? Sam Altman is prepared either way.
OpenAI CEO Sam Altman warns that investors are overexcited about AI, predicting someone will lose a "phenomenal amount of money" in what he compares to the dot-com bubble. This comes as OpenAI seeks a $500 billion valuation, up from $300 billion months earlier. Meanwhile, new MIT research shows 95% of enterprise AI pilots fail to deliver rapid revenue acceleration, though purchased AI tools succeed 67% of the time compared to internally built systems. The failures are attributed to implementation problems and learning gaps rather than AI model quality itself.
Towards Data Science·41wLangGraph + SciPy: Building an AI That Reads Documentation and Makes Decisions
A comprehensive tutorial on building an AI agent that helps users choose appropriate statistical tests by combining LangGraph for multi-step decision making with RAG (Retrieval-Augmented Generation) using SciPy documentation. The agent classifies user questions, searches embedded documentation when needed, provides recommendations, and generates sample Python code. The implementation includes ChromaDB for vector storage, OpenAI GPT-4 for language processing, and a Streamlit frontend for user interaction.
TechCrunch·41wSam Altman, OpenAI will reportedly back a startup that takes on Musk’s Neuralink
Sam Altman is reportedly co-founding Merge Labs, a brain-computer interface startup valued at $850 million, with funding expected from OpenAI's ventures team. The company will compete directly with Elon Musk's Neuralink, which recently raised $600 million at a $9 billion valuation and is currently conducting trials with paralyzed patients. This development represents another front in the ongoing rivalry between Altman and Musk, who previously worked together at OpenAI before their relationship deteriorated.
Theo - t3․gg·39wOpenAI just dropped a Cursor competitor?
OpenAI launched Codex, a comprehensive AI coding platform that competes with Cursor by offering IDE extensions, cloud-to-local environment switching, GitHub code review integration, and a rebuilt CLI in Rust. The service works across multiple editors (VS Code, Cursor, WindSurf) and includes background agents that can file pull requests automatically. Testing shows it can successfully build complete applications like an AI image generator with database persistence, though the UI has various bugs and the cloud-to-PR workflow has reliability issues. The platform offers different usage tiers ($20-200/month) with varying message limits, positioning itself as an all-in-one solution for developers already using ChatGPT.
gitconnected·42wHow I Built an AI-Powered Food Visualization Service in a Weekend
A developer built an AI-powered food visualization service in a weekend that converts menu text descriptions into realistic food images. The system uses a three-step pipeline: Tesseract.js extracts text from menu photos, OpenAI's GPT structures the raw text into dish descriptions, and Replicate's Stable Diffusion generates photorealistic food images. The full-stack application uses React/TypeScript for the frontend, Python Flask for the backend, and is containerized with Docker for easy deployment across platforms like Render, Railway, and Vercel.
Laravel News·41wThe Laravel Way to Build AI Agents That Actually Work
Vizra ADK is a new open-source Laravel framework that brings software engineering principles to AI agent development. It provides Laravel-style tools for building, testing, and deploying AI agents with features like automatic session management, tool calling, multi-LLM support, and a comprehensive evaluation system with 20+ built-in assertions. The framework includes Artisan commands, a web dashboard, API endpoints, and supports agent delegation and workflows, making AI development more structured and testable for Laravel developers.
Hacker News·42wopenai/harmony: Renderer for the harmony response format to be used with gpt-oss
OpenAI Harmony is a renderer library for the harmony response format used with gpt-oss models. The library provides consistent formatting for conversation structures, reasoning output, and function calls. It offers both Python and Rust implementations with fast performance through Rust core and Python bindings. The harmony format enables multi-channel outputs, tool calling, and structured responses that mimic OpenAI's API format.