Veda is an operating system that embeds an AI agent as a first-class system service rather than an add-on, making conversational control a native interaction channel alongside keyboard and mouse. The design argues that treating the agent as a peer of applications yields more reliable, visible and safe automation: every application exposes typed actions and structured state to the agent, actions run through the same code paths as manual input, and the system enforces consent by categorizing actions as routine, sensitive or destructive. Conversations start from real context (open windows, app state, time and explicit memory), voice is primary with a local wake detector and echo-cancelled audio, personal memory is stored locally and user-manageable, and Deepgram Voice Agent provides speech recognition and LLM services with configurable models and privacy settings.
The implementation demonstrates feasibility: everything is written in Rust from the UEFI bootloader through a capability-based microkernel (vkernel), user-space services, toolkit and applications, and it runs on x86-64 and in QEMU. Networking, TLS and crypto reuse mature crates (smoltcp, rustls, RustCrypto); the system uses a Linux driver VM under a paravirtual hypervisor with IOMMU isolation for third-party drivers; GPU compositing uses OpenGL ES on supported hardware; audio mixing, echo cancellation and ducking are built in; built-in apps expose over 70 actions. The agentapp protocol, structured consent UI, supervised services and capability IPC form a complete stack showing how an agentic-native OS can operate securely and transparently.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
Satya Nadella warns that all AI models should be considered 'compromised' and calls for an emergency halt on AI development. He emphasizes caution due to potential security risks and the need for stricter oversight. (theverge.com)
AI agents decompiled a popular first-person shooter into accurate, readable C++ code, focusing on behavior reproduction and portability. The project involved orchestrating multiple AI models, managing progress via GitHub, and enabling communication through Discord. (momo5502.com)
Littleguys is an open-source macOS tool that manages AI agents like Claude Code and Codex in the menu bar, providing real-time status and control. It allows users to chat, approve, schedule, and monitor agents across multiple machines without cloud dependencies. (littleguys.hervalicio.us)
Anthropic has issued a request for users to be less mean to its AI assistant, Claude, as part of a recent usage policy update. The company emphasizes respectful interactions with the AI to improve user experience. (theregister.com)
Anthropic's new policy bans extreme cruelty toward its Claude AI models, effective November 12, 2026, with enforcement using a conversation-ending tool. The policy clarifies that typical frustration or creative themes are not covered, but it does not specify what protections the cruelty ban aims to provide. (rews.cc)
Decision models can constrain language model outputs to fixed options by masking tokens and selecting the highest probability. This approach improves response reliability but does not guarantee correctness, and confidence scores reflect next-token prediction rather than true accuracy. (nishtahir.com)
Today's best Hacker News stories, summarized and screenshotted, one email a day.