AI Briefing
GeekNews news
The latest AI and developer news from GeekNews, with short summaries.
Visit website →National Supercomputer No. 6 'Han River' Fully Operational... Supporting Solutions to National Challenges such as K-Moonshot
The National Supercomputer No. 6 'Han River' has been fully operational, supporting AI-based science and technology innovation to solve national challenges such as K-Moonshot.
OpenQodex: Open-Source AI Code Review Tool Verifies Agent Changes with 98% Issue Recall
The npm package uses separate Claude Code or Codex agents to verify coding agent changes before push, achieving 98.0% issue recall in benchmarks.
Five Questions for Developers Collaborating with AI: Direction, Independent Execution, Verification, Delegation, and Learning
The improved speed of AI implementation has increased the importance of determining task direction, structure, verification, and the scope of delegation.
CRaC Reduces JVM Startup Time from 330 Seconds to 3 Seconds
Using Coordinated Restore at Checkpoint, the team enabled real autoscaling and reduced steady-state CPU usage to 20% of baseline.
Jepsen Finds 96 Consistency Issues in ParticleDB 1.1.4
Jepsen tests revealed 96 issues in ParticleDB versions 1.1.4 through 1.1.18.11.rc-22, including data loss and consistency violations in single-node setups, leading Jepsen to deem its 'Full ACID' claims premature.
PickEx-Meta Researcher Admits Torrenting Pirated Books3 Dataset, Deleted It Due to Litigation
Deposition testimony reveals a former Meta researcher downloaded the pirated Books3 dataset, contradicting Meta's earlier claim that it did not obtain the data via BitTorrent.
PickHerdr releases native desktop app for managing terminals and coding agents
The open-source tool provides a GUI for organizing AI agent workspaces, tracking agent status, and keeping terminals running in the background across local and remote machines.
OpenAI doubles usage limits for Pro and Business Premium tiers
Effective October 29, 2026, usage caps for Pro 100/200/500 and Business Premium plans are doubled, while Pro 200 Work and Codex limits remain unchanged.
PickReal-time voice booking bot built without a generative LLM, achieving 300–500ms latency using jev
Introduces a real-time voice booking bot demo that combines the non-generative model jev with code-based logic to address the latency and hallucination issues of generative LLMs.
Ferryx: Source-Available Native Terminal for AI Coding Agents on Windows, Linux, and macOS
The source-available tool supports parallel sessions for Claude Code, Codex, and other CLI agents with a headless Rust daemon that keeps processes running after the app closes.
kern: Daemonless rootless container runtime for AI code execution
The tool claims 80x faster startup than Docker for single containers and supports Python/Node SDKs for sandboxing LLM-generated code.
Judge rules Anthropic illegally trained Claude on pirated books
A federal judge ruled that Anthropic illegally scraped and pirated works from over 10 Stanford professors to train its Claude models, despite a $1.5 billion settlement.
Claude Code skill converts arXiv papers to translated books preserving LaTeX structure
The translate-book-arxiv agent skill reads original LaTeX sources to maintain equations and numbering while outputting PDF, DOCX, and EPUB files in seven languages.
Spring AI Privacy Guardrails adds PII tokenization and tool authorization
The library tokenizes PII before model calls and restores it only for authorized tools, supporting Spring AI 2.0.0 and Spring Boot 4.0.0.
Roblox crawler downloads 145GB of 3D models including adult content from Open3DLab
Operators of Open3DLab, SFMLab, and SmutBase reported that a Roblox-affiliated crawler bypassed robots.txt and terms of service to scrape 1,027 files totaling 145.5GB, with 62.6GB comprising adult content.
PaperPilot: Pure Rust PDF engine with WASM and MCP support
The open-source tool runs locally via CLI, browser, or desktop app without cloud calls, supporting PDF merge, split, OCR, and compression.
Kimi K3.1 Pro Scores 55 on Artificial Analysis Intelligence Index
The 2.8 trillion parameter model ranks 4th out of 215 models on the index but is noted for slow output speed and high verbosity.
Sphinx: Git hook tool quizzes developers on AI-generated code before push
Sphinx intercepts git push to protected branches and requires users to answer LLM-generated questions about code changes to verify understanding.
LambdaDB Achieves 95% Input Token Savings in AI Memory Pipeline Using Native Embeddings
The pipeline reduces average input tokens from 105,183 to 4,886 while maintaining 96.0% accuracy on LongMemEval, outperforming full-history baselines.
Pickgpu-unlock Linux module doubles FP/BF16 instruction rate on Nvidia 3000 series GPUs
The open-source tool unlocks hardware limits on GA102-GA106 chips, boosting BF16 GEMM throughput by up to 81% and LLM prefill speeds by 14-36%.