Anthropic has launched an auto mode for its Claude Code tool, allowing the AI to act on users' behalf while reducing the risk of unwanted actions. The feature flags and blocks potentially risky operations, prompting the model to retry or request user intervention. Currently available as a research preview for Team plan users, Anthropic plans to extend access to Enterprise and API users in the coming days. The company emphasizes that the tool remains experimental and recommends use in isolated environments. Read more →

OpenAI announced a new set of open‑source prompts designed to help developers build AI applications that are safer for teenagers. The prompts address a range of risky content, including graphic violence, sexual material, harmful body ideals, dangerous challenges, and age‑restricted services. By providing clear, operational safety policies, OpenAI aims to give developers a practical foundation for protecting younger users, while acknowledging that the broader challenges of AI safety remain complex. Read more →

Anthropic has introduced Claude Cowork, a research‑preview AI assistant for Claude Pro and Max subscribers that can perform computer tasks on macOS and Windows without complex setup. The tool can open files, browse the web, interact with apps, and run developer utilities, using built‑in connectors for services like Gmail, Google Drive, and Slack when available, and otherwise controlling the mouse and keyboard. It always requests permission before accessing new apps or files and can be stopped at any time. Additional features include Claude Dispatch for mobile command input, Claude Channels for event integration, and scheduled task automation. Read more →

Cursor introduced its new coding model, Composer 2, touting frontier-level coding intelligence. An X user quickly flagged that Composer 2 is essentially a rebranded version of Moonshot AI’s open‑source Kimi 2.5, with additional reinforcement learning. Cursor’s vice president of developer education confirmed that the model started from an open‑source base, noting that only a portion of the compute used for the final model came from the original Kimi code. The company defended the use as compliant with Kimi’s license and described the partnership with Moonshot AI as authorized and collaborative. Read more →

OpenAI has launched two smaller versions of its latest GPT-5.4 model—Mini and Nano—designed for developers who prioritize speed and cost over maximum reasoning power. The Mini model runs more than twice as fast as the full model while staying close on key benchmarks, and the Nano model focuses on simple classification and data‑extraction tasks. Both models support text and image inputs, tool use, function calling, and a 400,000‑token context window, and they are available today via the API, Codex, and ChatGPT. This tiered approach lets developers allocate cheaper models for routine work and reserve the full model for complex reasoning, reshaping how real‑time AI applications are built. Read more →

Y Combinator CEO Garry Tan unveiled an open‑source Claude Code configuration called gstack, sharing it on GitHub under an MIT license. The project quickly amassed thousands of stars and forks, drawing enthusiastic support on platforms like Product Hunt. At the same time, the release provoked criticism from developers who dismissed it as merely a collection of prompts and questioned its novelty. Expert AI models, including Claude, ChatGPT and Gemini, offered largely positive assessments, describing gstack as a sophisticated prompt workflow. The mixed reaction highlights both excitement and skepticism surrounding AI‑augmented coding tools. Read more →