OpenAI retired several older ChatGPT consumer models, including GPT-4o, immediately upon GPT-5 launch to simplify the user experience and reduce confusion from the prior model picker. Conversations using retired models automatically switched to the closest GPT-5 variant or specialized modes like GPT-5-Thinking. This removal frustrated users who preferred GPT-4o for creative, emotional, and long-form interactions rather than coding tasks. OpenAI CEO Sam Altman responded by promising to restore GPT-4o access for Plus subscribers, reflecting openness to community feedback. The API retains access to all legacy models. GPT-5 also introduces safer, more cautious response behaviors to handle delicate personal topics, raising ethical tensions between model safety and diverse user needs.

The flip-card project reimagines a business card as a microcontroller-driven device running a fluid-implicit-particle (FLIP) simulation, showcasing advanced PCB design and numerical fluid simulation on an RP2350. Key technical elements include open KiCad PCB files, custom rechargeable battery with a fragile board-edge USB-C charger, and a WASM-based simulator for development. The project balances artistic fluid dynamics with practical challenges like text legibility on small LED grids and device thickness. Community discussions focus on enhancing interactivity via accelerometer taps, adding QR codes, and careful tradeoffs in hardware fragility versus functionality. The flip-card demonstrates the intersection of embedded programming, physics algorithms, and hardware aesthetics in a compact form.

Manish documents building a fully localized AI workspace emphasizing privacy by eliminating cloud dependencies. The system runs large language models, executes generated code within sandboxed lightweight VMs (via Apple’s container tool), and automates web research using a headless browser. Key components include Ollama for local LLMs, coderunner for orchestration, and Playwright for browser automation. Despite integration challenges—such as incomplete tool support, platform-specific quirks, and Mac app development difficulties—the setup successfully performs tasks like video editing, image manipulation, and software installation without exposing code or data externally. The project appeals to developers interested in private AI tooling, container isolation, and local LLM deployment.

Ben Collier traces Tor’s evolution from a Cold War U.S. Navy project developing onion routing to obscure metadata by layered encryption, into a global privacy tool relied on by activists, journalists, and dissidents. The article emphasizes an unlikely alliance between cypherpunks and military researchers united by a vision of privacy as a structural property embedded in digital systems. Tor’s open and widespread design protects anonymity by blending risky users into normal internet traffic patterns. The piece critiques modern legislative attempts, like the UK’s Online Safety Act, to undermine encryption—arguing that dismantling privacy tools harms vulnerable communities by increasing surveillance rather than protecting them. It highlights the importance of community-led moderation over heavy centralized control.