AI news

What Is Copilot Exactly? (idiallo.com) AI

The article explains that “Copilot” can refer to several different Microsoft AI products (for example, GitHub Copilot, Copilot for Microsoft 365, Windows Copilot, and Copilot Chat), each integrated into different tools and workflows. The author shares a week-long attempt to improve their productivity with Copilot for Teams/Microsoft 365 before realizing others may be using a different “Copilot” entirely. It ultimately frames the confusion as a caution to clarify which specific tool people mean when they say they use “Copilot.”

The OpenAI Graveyard: All the Deals and Products That Haven't Happened (forbes.com) AI

The article reviews major OpenAI deals and product announcements that, despite prior hype, did not end up materializing, mapping out what was promised versus what was never launched or only partially delivered.

Show HN: Real-time dashboard for Claude Code agent teams (github.com) AI

Show HN introduces agents-observe, a GitHub project that provides a real-time observability dashboard for Claude Code and multi-agent sessions. It uses Claude Code “hooks” to stream tool calls, subagent lifecycles, and file/tool activity into a local or remote server that stores events in SQLite and pushes updates over WebSockets to a React UI. The dashboard supports filtering/searching across agent events and viewing the agent hierarchy to make autonomous debugging less dependent on post-hoc logs.

StepFun 3.5 Flash is #1 cost-effective model for OpenClaw tasks (300 battles) (app.uniclaw.ai) AI

The UniClaw OpenClaw Arena cost-effectiveness rankings report that StepFun 3.5 Flash is currently ranked as the most cost-effective model for OpenClaw tasks, based on 300 battles.

Apple Removes iPhone Vibe Coding App from App Store (gizmodo.com) AI

Apple removed the “Anything” iPhone app from the App Store, citing a violation of App Store Guideline 2.5.2 about apps being self-contained and not downloading, installing, or executing code that changes features or functionality. The move follows earlier blocks of “vibe coding” apps such as Replit and Vibecode, which use AI assistance to generate or modify other apps. Apple did not immediately provide details to Gizmodo, while Anything’s CEO says attempts to adjust the app were rejected and that the enforcement appears to be tightening around this category.

We Built It with Slide Rules. Then We Forgot How (unmitigatedrisk.com) AI

The post argues that spaceflight know-how—once built through hands-on experimentation and then preserved in documents like NASA SP-287—has been eroding as organizations grow too complex and stop asking basic operational questions. It recounts the author’s father learning rocket chemistry and working on satellite attitude control, then contrasts that transferable “keep it in your head” approach with modern Artemis planning, which the author says reflects hidden constraints and insufficient familiarity among leaders. The author extends the warning to software and AI, suggesting capability can be outsourced before judgment and underlying understanding are transmitted, leaving teams “renting” complexity without owning the decisions.

I Quit. The Clankers Won (dbushell.com) AI

The author argues that despite claims that blogging is “over,” now is a crucial time to keep writing to preserve authentic human voices in an industry increasingly dominated by AI hype, plagiarism machines, and surveillance. They also criticize generative AI (including Sora) as largely low-value “slop,” and encourage readers to avoid Big Tech narratives and use blogging to support an open, indie web.

AI has suddenly become more useful to open-source developers (zdnet.com) AI

ZDNET reports that open-source maintainers are increasingly finding AI coding and security tools more reliable for real-world tasks, improving report quality and helping with legacy code maintenance. The article also highlights ongoing concerns, including potential legal disputes over AI-assisted rewrites, and the flood of low-quality “AI slop” that can overwhelm projects. Organizations like OpenSSF are working to make better AI tools available to maintainers as reliability continues to improve.

Show HN: Baton – A desktop app for developing with AI agents (getbaton.dev) AI

Baton is a desktop app for running AI coding agents with separate, git-isolated workspaces so multiple agents can work in parallel without stepping on each other. It provides a dashboard to monitor agent status, view diffs and file changes, manage worktrees, and open pull requests from the app, while running CLI agents in real terminal sessions. The project claims code stays local, with optional AI-generated workspace titles/branch names handled via a paid provider and supporting custom or first-class integrations like Claude Code, Codex, and others.

OpenAI closes funding round at an $852B valuation (cnbc.com) AI

OpenAI has closed a record $122 billion funding round at a post-money valuation of $852 billion, up from $110 billion previously announced. The round was co-led by SoftBank and included investors such as Andreessen Horowitz and D. E. Shaw Ventures, and OpenAI also added participation via bank channels plus $3 billion from individual investors. The company is not yet profitable and continues to burn cash as it prepares for potential IPO scrutiny.

Show HN: 1-Bit Bonsai, the First Commercially Viable 1-Bit LLMs (prismml.com) AI

PrismML announces “1-bit Bonsai” models that use 1-bit weights to shrink memory and power requirements for running LLMs on edge devices and in robotics. The company claims the 8B model fits in about 1.15GB of RAM, runs faster and more energy-efficiently than full-precision 8B models, and preserves benchmark performance. It also offers smaller 4B and 1.7B variants designed for on-device speed, with detailed comparisons reportedly covered in a whitepaper.

TinyLoRA – Learning to Reason in 13 Parameters (arxiv.org) AI

The paper introduces TinyLoRA, a parameter-efficient adapter method that scales reasoning performance using extremely small low-rank updates (as few as 13 trained parameters). The authors report that training an 8B Qwen2.5 model with TinyLoRA reaches about 91% accuracy on GSM8K and recovers roughly 90% of performance gains on harder reasoning benchmarks while using 1,000× fewer parameters than typical approaches. They also find the strong results depend on reinforcement learning, with supervised fine-tuning requiring much larger updates to match performance.

Claude Code Unpacked : A visual guide (ccunpacked.dev) AI

Claude Code Unpacked is a visual, source-based guide that walks through how Claude Code works, from user input and an agent “loop” to rendering responses, tool execution, and command handling. It catalogs Claude Code’s built-in tools, slash commands, and optional/hidden features (including unreleased or feature-flagged capabilities), with links to the relevant parts of the codebase. The site is unofficial and notes that some details may be outdated or inaccurate.

President's new science council: 9 billionaires and 1 scientist (scientificamerican.com) AI

U.S. President Donald Trump has named a new PCAST science and technology advisory council dominated by technology leaders, with 9 billionaires and only one university researcher, quantum physicist John Martinis. The panel is largely focused on areas like artificial intelligence, quantum information, and nuclear power, and critics say it lacks representation from biology and broader academic expertise. The administration could add up to 11 more members under a 2025 order.

The Claude Code Source Leak: fake tools, frustration regexes, undercover mode (alex000kim.com) AI

A blog post says Anthropic accidentally exposed the full, readable source code of its Claude Code CLI via an npm source map leak that was quickly mirrored after the package was pulled. The author describes several built-in mechanisms, including server-side “anti-distillation” with fake tool injection, an “undercover mode” that can hide an AI’s internal identifiers in external repos, and regex-based detection of user frustration. The post also notes client attestation logic intended to verify official binaries, product code references to a feature-gated autonomous agent mode, and commentary that the leak comes shortly after related legal disputes over third-party API use.

TSMC is reportedly sold out until 2028 (pcgamer.com) AI

TSMC is reportedly booked through 2028 for its N2 process, with even some future capacity at not-yet-built plants reportedly reserved. A South Korean report says reservations for TSMC’s planned Arizona Fab 4 (targeting mass production by 2030) are already closed and that additional demand—from both major chip customers and AI-driven firms—may push buyers to consider alternatives like Samsung. The article argues that this lack of available leading-edge foundry capacity could keep prices and supply for advanced GPUs and CPUs constrained for years.

Raspberry Pi profit surges as AI boom lifts demand (ft.com) AI

The article says Raspberry Pi’s profits have risen sharply as demand increases tied to the wider AI boom, boosting interest in the company’s small computing hardware. It links the company’s improved financial performance to a surge in orders for AI-related and related development use cases.

Someone just converted Claude Leark from TypeScript to 100% Python (github.com) AI

A GitHub project, instructkr/claw-code, describes a clean-room rewrite of the “Claude Code” agent harness, moving the active codebase to Python (and noting a separate Rust port in progress). The repository’s README explains why the leaked snapshot is no longer tracked as the main source, outlines the current Python workspace structure, and provides commands for generating a manifest/summary and running tests or parity checks. The post also credits use of an AI-assisted workflow tool (oh-my-codex) and links to an accompanying discussion about legal/ethical issues.

Project Mario: the inside story of DeepMind (colossus.com) AI

An excerpt from Sebastian Mallaby’s book describes how DeepMind co-founders Demis Hassabis and Mustafa Suleyman tried to build AI safety governance inside Google, beginning after a failed 2015 oversight board meeting involving Elon Musk. Their “Project Mario” talks with Google and Alphabet aimed to create a semi-independent structure with a 3-3-3 board, but internal resistance from Google leadership derailed a hoped-for spin-out and pushed them toward a potential $5 billion outside-investor “walk away” plan framed as serving the public interest.

Accidentally created my first fork bomb with Claude Code (droppedasbaby.com) AI

A software engineer recounts how an agentic “hook” in Claude Code recursively spawned new Claude Code instances, effectively creating an accidental fork bomb that overheated and froze their Mac overnight. After quickly removing the hook and preventing further runaway processes, they report it also likely avoided a much larger corporate API bill—though they’d already seen it spike by hundreds of dollars. The post then describes the practical custom tools and skills they built for everyday workflow (e.g., task triage, OCR, local memory/metadata logging), despite the costly experiment.

From 300KB to 69KB per Token: How LLM Architectures Solve the KV Cache Problem (news.future-shock.ai) AI

The article explains how a transformer’s KV cache makes ongoing conversations “remember” recent tokens in GPU memory, and why its byte cost forces constant memory management. It compares several architecture changes—like grouped-query attention, compressed latent caches, and sliding-window attention—that reduce per-token cache size, and contrasts this short-lived working memory with long-term “memory” features that rely on separate systems such as retrieval and stored facts. It also discusses what happens when the cache is evicted or too large, including lossy compaction and the resulting need for external memory tools.

Microsoft: Copilot is for entertainment purposes only (microsoft.com) AI

Microsoft’s Copilot terms of use outline how the AI service may be accessed and what rules users must follow, including requirements around age, lawful personal use, and a broad code of conduct (privacy, non-harm, no fraud, no deepfakes, etc.). The policy also warns that Copilot can make mistakes and may use unreliable or unverified information, advising users to rely on their own judgment. Microsoft further states Copilot is “for entertainment purposes only” and that users shouldn’t depend on it for important advice, while additional provisions address possible access restrictions and third-party “shopping” handled by merchants.

Good code will still win (greptile.com) AI

Greptile’s blog argues that even as AI coding accelerates, “good code” will ultimately prevail because maintainable, simple code is cheaper to generate and fix over time. It points to trends like larger, denser pull requests and increasing outages as signs that brute-force coding can make systems more brittle. The piece suggests market competition and the economics of long-term maintenance will push AI tools toward clearer abstractions and fewer changes rather than “slopware.”

Lime (bikes) is a data company (ktoya.me) AI

An author uses a GDPR data request to obtain three years of their Lime bike history, then analyzes the trip and app logs with Claude to build dashboards and identify patterns. The analysis includes their spend, ride frequency, and loyalty segmentation, and it also infers likely home/work locations and routine stopovers (e.g., gym, brunch, a regular Tuesday appointment) from GPS timing. The post argues that similar approaches can be applied to other EU/UK consumer apps that store data, using an AI agent to explore and visualize it.

Cohere Transcribe: Speech Recognition (cohere.com) AI

Cohere has released Transcribe, an open-source, conformer-based automatic speech recognition (ASR) model trained to minimize word error rate while remaining production-ready. The 2B-parameter model supports 14 languages and is reported to rank #1 on Hugging Face’s Open ASR Leaderboard for English accuracy (5.42% average WER), with similar gains claimed in human evaluations. Cohere says it also delivers strong throughput and is available for local use, via a free API for experimentation, or through its Model Vault for managed, low-latency deployment.