Read what matters
in Tech & AI.

Andrew Ferguson of the F.T.C. said in a letter to Apple that it might be violating consumer protection law by stifling conservative speech in its news aggregation service.

Discussion | Link

Ben Thompson's Stratechery highlights its recent content, including a video analysis of Microsoft's strategic positioning in the ever-evolving software landscape. This week's focus centers on Microsoft's resilience and adaptation, particularly concerning its approach to artificial intelligence and competition.

Discussion | Link

Large Language Models (LLMs) are rapidly changing the software development landscape, impacting roles from junior developers to seasoned architects. Insights from the Thoughtworks Future of Software Development Retreat reveal how LLMs are prompting a re-evaluation of developer responsibilities, team structures, and the very nature of cognitive workload in the age of AI.

In Februrary 2026, Thoughtworks hosted a workshop called “The Future of Software Development” in Deer Valley Utah. While it was held in the mountains of Utah as a nod to the 25th anniversary of the writing of Manifesto for Agile Software Development, it was a forward-looking event, focusing on how the rise of AI and LLMs would affect our profession. About 50 or so people were invited, a mixture of Thoughtworkers, software pundits, and clients - all picked for being active in the LLM-fu

This week's 'Hard Fork' podcast dives into the anxieties surrounding AI-driven job displacement sparked by Matt Shumer's viral essay, explores how AI is revolutionizing (or perhaps homogenizing) the romance novel industry, and highlights positive tech advancements. Hosts Kevin Roose and Casey Newton navigate the complex landscape of AI's impact on the economy and creative fields, offering both cautionary tales and glimpses of technological progress.

JDoodle.ai has launched MCP, a groundbreaking platform that allows developers to build and deploy web applications directly from AI chatbot interfaces like ChatGPT and Claude. This innovative approach promises to significantly accelerate the development lifecycle and democratize access to web app creation. The platform is currently generating buzz on Product Hunt.

A new preprint shows GPT-5.2 proposing a new formula for a gluon amplitude, later formally proved and verified by OpenAI and academic collaborators.

The Iranian government is escalating its use of digital surveillance to suppress dissent, employing tactics such as location tracking, facial recognition, and communication blackouts to identify and target protesters. Human rights groups and researchers warn that this sophisticated surveillance infrastructure poses a significant threat to civil liberties and freedom of expression within the country. These efforts come in response to recent anti-government demonstrations.

OpenAI is rolling out new security features for ChatGPT, aimed at protecting organizations from sophisticated prompt injection attacks and AI-powered data exfiltration. Lockdown Mode and Elevated Risk labels provide crucial safeguards against emerging threats that exploit large language models (LLMs). This proactive approach aims to ensure the responsible and secure use of AI in enterprise settings.

Discussion | Link

OpenAI has engineered a sophisticated access management system that blends rate limits, granular usage tracking, and credit allocation to ensure continuous availability of its high-demand AI models, including Sora and Codex. This architecture is crucial for democratizing access while preventing resource exhaustion and maintaining service quality. The innovative system provides a blueprint for managing access to computationally expensive AI tools.

OpenAI has unveiled GABRIEL, a groundbreaking open-source toolkit designed to transform qualitative data, like text and images, into quantifiable insights for social scientists. By leveraging the power of GPT models, GABRIEL promises to unlock new possibilities for large-scale research and analysis in the field.

Discussion | Link

Discussion | Link

OpenAI’s Sherwin Wu on why the productivity gap is widening, how his engineers now manage 10-20 parallel agents, and why you need to build for where models are heading

An interview with me by John Collison on the Cheeky Pint podcast about AI, ads, and the history of Stratechery.

HookWatch emerges as a dedicated webhook monitoring solution tailored for indie hackers and small teams. This new tool promises to streamline the process of ensuring webhook reliability, providing crucial insights into delivery status and potential failures. Is this the key to hassle-free integrations?

Discussion | Link

Discussion | Link

A new era of coding assistance is dawning! GPT-5.3-Codex-Spark, a real-time coding model boasting 15x faster generation speeds and a massive 128k context window, has landed in research preview for ChatGPT Pro subscribers. This breakthrough promises to transform the development workflow, offering unprecedented speed and understanding for complex coding tasks.

Imagine an AI that speaks with your unique vocal style. Ghostype, a novel AI voice interface, promises to learn and replicate your individual speech patterns, opening up new possibilities for personalized digital communication and content creation.

Discussion | Link

Discussion | Link

Discussion | Link

A software engineer recently pitted Anthropic's Claude Opus 4.6 against OpenAI's GPT-5.3 Codex in a real-world coding challenge, dramatically accelerating their workflow. The results highlight the strengths and weaknesses of each AI model, revealing how they can be strategically combined to revolutionize software development. This experiment provides valuable insights into the future of AI-assisted programming.

Spotify's unique position as a content network allows it to harness the power of AI for continued growth and innovation. With a strong business model in place, the platform is poised to further personalize user experiences and optimize content delivery through sophisticated AI algorithms. This approach strengthens Spotify's competitive edge in the ever-evolving music streaming landscape.

Discussion | Link

The rise of AI agents is upon us, and a critical factor in their success is how effectively we can 'harness' the power of large language models (LLMs) like Codex. A new approach, termed 'harness engineering,' focuses on optimizing prompts and environments to unlock the full potential of these models, leading to more robust and capable AI agents. This could redefine software development and automation as we know it.