× + newsletter-2026-03-26
╔═════════════════════════════════════════════════════════════╗
║                                                             ║
║   ▀█▀ ▄▀█ █▄▀ █▀▀   █ ▀█▀   █▀▀ █▀█ █▀█ █▀▄▀█             ║
║    █  █▀█ █ █ ██▄   █  █    █▀  █▀▄ █▄█ █ ▀ █             ║
║                                                             ║
║   ▀█▀ █ █ █▀▀   ▀█▀ █▀█ █▀█                                ║
║    █  █▀█ ██▄    █  █▄█ █▀▀                                ║
║                                                             ║
╚═════════════════════════════════════════════════════════════╝
    

// Tech news at terminal velocity

cat 2026-03-26.md

OpenAI's Cold Shower, Dystopian Police Drones, and Copilot's Data Grab

$ cat TLDR.md

• OpenAI indefinitely shelves its highly anticipated (and controversial) 'adult mode' for ChatGPT.

• GitHub Copilot updates its policy to train on your interaction data by default starting April 24th.

• Uber brings robotaxis to Europe, starting in Croatia with Pony.ai and Verne.

Headlines & Launches

💔 OpenAI's Erotic Chatbot Gets the Cold Shower

OpenAI has indefinitely paused its plans for a sexualized 'adult mode' for ChatGPT. After facing pushback from employees and investors over the problematic nature of the project, the company is refocusing on its core products. Sorry, you'll have to find romance elsewhere.

Source: The Verge

🚕 Uber's Croatian Robotaxi Adventure

Uber is teaming up with Chinese autonomous driving firm Pony.ai and Croatian startup Verne to launch Europe's first commercial robotaxi service. The fleet will hit the streets of Zagreb, Croatia, before expanding to other European cities.

Source: Reuters

🎸 Google Drops Lyria 3 Pro for AI Musicians

Google just released Lyria 3 Pro, a music generation model that can pump out structurally aware, three-minute tracks. You can now prompt for specific intros, choruses, and bridges via the Gemini API and Vertex AI.

Source: TechCrunch

🌪️ Mistral's 'Small' Model Does It All

Mistral just dropped Mistral Small 4, a 119B parameter Mixture-of-Experts model that handles chat, reasoning, and coding all from a single endpoint. They also released new open-source speech generation models that are efficient enough to run locally on a smartwatch.

Source: Analytics Vidhya

Deep Dives

🕵️‍♂️ GitHub Copilot Wants Your Data

Starting April 24th, GitHub Copilot will begin using interaction data—including inputs, outputs, and code snippets—from Free, Pro, and Pro+ users to train its AI models. If you don't want your late-night spaghetti code becoming part of the hive mind, you'll need to manually opt out in your settings.

Source: GitHub Blog

🍿 The LibreOffice Donation Banner Drama

LibreOffice is planning to add a modest donation banner to its Start Center in version 26.8, and the open-source community is predictably losing its collective mind. It's a fascinating look at the inherent tension between providing free software and the harsh reality of needing money to keep the lights on.

Source: Document Foundation

Engineering & Research

🕷️ Robust LLM Extractor for Websites

Tired of rewriting CSS selectors every time a website changes its layout? Lightfeed has open-sourced a TypeScript-based LLM extractor that uses AI to reliably scrape and structure data from web pages, even when the DOM shifts.

Source: GitHub

🦅 Swift 6.3 is Officially Here

Apple has released Swift 6.3, bringing new features and refinements to the language. It's a solid update for Apple ecosystem developers looking to modernize their codebases and improve performance.

Source: Swift.org

🛡️ Delve Secures LiteLLM After Malware Hit

After the popular open-source AI project LiteLLM was infected by credential-harvesting malware, security firm Delve stepped in to handle the compliance and cleanup. It's a stark reminder to audit your AI supply chain before it audits you.

Source: TechCrunch

Odds & Ends

🚁 The Dystopian Police Drone Has Arrived

BRINC's new 'Guardian' police drone comes equipped with Starlink, a 130dB siren, and the ability to chase vehicles at 60mph. On the bright side, it can also drop Narcan and defibrillators, making it the most aggressively helpful robot in the sky.

Source: Ars Technica

👽 Reddit Demands Proof of Humanity

Reddit is rolling out a new policy requiring 'fishy' accounts to verify they are actually run by a human. Ironically, AI-generated content is still perfectly acceptable, as long as a carbon-based lifeform clicked the 'post' button.

Source: Ars Technica

← Back to archive