OpenAI has released ChatGPT Images 2.5, which includes a new Sketch feature allowing users to draw a rough doodle inside ChatGPT and have it converted into a polished image based on typed instructions. Users can invoke the tool by typing @Sketch, then annotate specific parts of the generated image with comments to request targeted edits, such as changing eye color. OpenAI says the update also improves lighting and texture quality, strengthens multi-turn editing consistency, and cuts image generation latency by up to 50% compared to the previous version.
A developer released Solid Objects, an open-source library that replicates the Durable Objects programming model—one identity, one state, one ordered mailbox per object—without requiring a dedicated vendor runtime or daemon. Instead, it runs directly on databases developers already operate, including SQLite, PostgreSQL, and MySQL. The release follows the creator's frustration with vendor lock-in and unpredictable metered billing tied to existing Durable Objects platforms.
A design educator who teaches a course on behavior change argues that every design decision—whether for a book cover, app, or ad—is ultimately an attempt to influence user actions, most often to boost sales or shareholder value. The essay contrasts public companies, B corps, and nonprofits, noting that even mission-driven organizations use design to drive fundraising or other outcomes, and that designers rarely know an employer's true motives until they're inside.
An international team of scientists published a not-yet-peer-reviewed paper arguing that large language models like ChatGPT and Claude spread through populations similarly to how viruses transmit, driven by social learning and usefulness rather than biology. They contend this spread encourages 'cognitive offloading,' where people increasingly hand over mental tasks like reasoning and writing to AI tools, deepening reliance over time.
Meta introduced Muse, a personal AI agent that can autonomously handle tasks like online shopping, sending emails, and trip planning by opening browsers, filling out forms, and even negotiating on a user's behalf. The assistant will launch in the US on iOS, Android, and muse.ai, with support for AI glasses planned, and will be free for most users while Meta develops unspecified paid tiers for heavier use.
Tom's Hardware benchmarked Alibaba's newly released Qwen3.8 27B open-weight AI model across discrete GPUs like the RTX 5090, RTX 4090, RTX 3090, and AMD/Intel cards, as well as unified-memory systems including the DGX Spark, Mac Studio, and Ryzen AI Halo. Despite the four-bit quantized model needing only about 17GB of VRAM, the tests found that software stacks and inference engines often bottleneck real-world performance, affecting time-to-first-token and throughput once context windows fill up.
A class action lawsuit filed by former FTC attorneys Monica Vaca and Kati Daffan accuses Anthropic of misleading Claude Max subscribers about how much extra usage their $100 or $200 monthly plans actually provide. The suit claims Anthropic's '5x' and '20x' usage claims, prominently advertised, are undercut by fine print restricting those multipliers to rolling five-hour windows capped by weekly limits, rather than a straightforward overall usage boost.
Copperhead is a new tool positioned as bringing software-like speed to hardware development, released as open core. Its command-line interface is free to use and performs the core functionality independently, while paid tiers cover hosting, team collaboration, and audit-ready features.
A team of developers has completed a full decompilation of Super Smash Bros Melee, converting Nintendo's 2001 GameCube fighter into readable C code that compiles into a binary identical to the original main.dol. The project, hosted on the doldecomp/melee GitHub page, took over six years and reportedly used OpenAI's Astra to help finish the work.
OpenAI confirmed an ongoing outage causing errors in ChatGPT's image generation and its Images API, along with failures and delays when users try to upload files. Hundreds of users reported the problems on DownDetector, and OpenAI says it is still assessing mitigation efforts as some requests continue to fail.
A technical blog post walks through how the bitap (shift-and) algorithm finds a pattern inside a text, building it up step by step from the naive brute-force matching approach. The author shows how the method can be adapted to process text as a stream of characters rather than requiring the whole string in memory, and highlights its use of bitwise operations for efficiency when the search pattern is shorter than a machine word.
A modder known as ShyVox released 'DLSS Unlocked,' which merges the OptiScaler_DLSSNR and DLSS Enabler projects into one installer that adds DLSS 5 and multi-frame generation support to RTX 20-, 30-, and 40-series cards. On RTX 20 and 30 GPUs, AMD's FSR 3.1 handles frame interpolation since Nvidia's own pipeline isn't supported, while RTX 40 cards use Nvidia's DLSS-G for the first generated frame and FSR 3.1 for the rest, enabling up to 6x frame multiplication.
GoKawiil