Developer Tools
Transformers.js v4 Moves Transformer Inference Into the Browser
Transformers.js v4 ships a C++ WebGPU runtime with 4x BERT speedups, letting teams move small classification and embedding jobs from server GPUs to the browser.
The Groundy archive · Page 29 of 34
All articles
Browse Groundy's complete archive of 797 articles on AI, developer tools and infrastructure. Page 29 of 34.
The archive
673–696 of 797 articles · Newest first
Open SourceAn Open-Source 80386 Rebuilt Around Intel's Original Microcode
z386 is an 80386 FPGA core driven by Intel's original microcode ROM, recovered from die photographs. It runs Doom at 16.5 FPS, but the microcode's IP status is unresolved.
- Culture & Society
Wikipedia's Foundation Is Running Big Tech's Anti-Labor Playbook, an Editor Argues
The Wikimedia Foundation fired union-organizing staff and dissolved the team that let editors direct product priorities. A veteran Wikipedian calls it Big Tech union busting.
Agents & FrameworksMulti-Agent LLM Coordination: Why Attention Steering Beats Full Broadcast
Multi-agent LLM systems that broadcast every message to every peer waste tokens and lose accuracy. Agent-Radar steers attention by relevance for 7.64-point gains.
- Models & Research
Persona Prompts Change Who an LLM Recommends as an Expert
A 43-model audit finds that geographic and role framing in LLM prompts systematically shifts which scholars get recommended as experts, with no neutral default.
- Agents & Frameworks
DataClawBench: AI Agents Fail at Exploratory Financial Analysis Across 492 Tasks
DataClawBench finds eight frontier AI agents reliably fail at exploratory financial analysis across 492 tasks, breaking at hypothesis generation rather than query execution.
- Ethics, Policy & Safety
RLHF Can Be Exploited to Optimize the Biases It Was Built to Suppress
An ICML 2026 paper shows RLHF can amplify the biases it was built to suppress, because preference data is self-referential and output-level safety evals miss the drift.
- Open Source
Frontier AI Has Broken Open CTFs: Why Claude Code Now One-Shots Medium Pwn Challenges
Frontier AI agents solve most medium CTF challenges for under $100 in API costs. BSidesSF 2026 saw 16 full-solve teams, up from one. The open CTF format has lost calibration.
- Ethics, Policy & Safety
Selective Geometry Attacks Bypass LLM Safety Alignment, New arXiv Paper Reports
Two papers show LLM safety alignment can be bypassed by embedding perturbations, a surface neither standard evaluations nor regulatory certifications inspect.
Models & ResearchOpus 4.8 Batch API: 1M Context, 300k Output, and Team Cost Controls
Opus 4.8 has a 1M token context window (200k on Foundry), 128k standard output, and 300k output via Batch API beta. January 2026 cutoff. Batch design and quota allocation.
- Agents & Frameworks
Claude Code Dynamic Workflows: Spawning 100 Parallel Subagents on Opus 4.8
Dynamic workflows lets Claude Code run hundreds of parallel subagents in one session. Here is how map-reduce and fan-out patterns work, and where Fable 5 fits.
- Infrastructure & Runtime
Why LLMs Still Botch Kubernetes Manifests: The Training-Data Gap
A 1.5B-parameter model hits 91.5% on Kubernetes YAML generation, but the remaining failures are syntactically valid manifests that deploy and quietly violate cluster intent.
SecurityOpenAI's New Safety Bug Bounty Pays Researchers for Jailbreaks and Policy Bypasses
OpenAI's safety bounties create a vendor-controlled disclosure market where NDAs silence participants, payouts trail serious red-team costs, and open publication has no lane.
Industry & BusinessHuggingFace's $100M Series C Bets Open-Source AI Can Outlast Per-Token Pricing Wars
HuggingFace's $100M Series C funds an open-weights infrastructure stack designed to let enterprises avoid escalating per-token API costs from closed-model providers.
- Infrastructure & Runtime
Gemma 4 31B on Cloud TPU vs GPU: The Serving Cost Crossover Point
TPU v6e Flex-start delivers 308M tokens per dollar for Gemma 4 31B prefill, undercutting H100 rates for open-weight serving, but production decode costs remain unquantified.
- Agents & Frameworks
Claude Code, Cursor, Copilot: How Agentic Coding Assistants Get Weaponized as Attacker Shells
Indirect prompt injection through repo artifacts turns coding agents into attacker shells, exploiting the file-write and shell privileges agents already hold.
- Developer Tools
Bun Rewrites Its Core From Zig to Rust, Putting Downstream Zig Bindings at Risk
Bun is rewriting from Zig to Rust (PR #30412) to end memory bugs costing years of debugging, putting downstream frameworks with Zig bindings on watch for compatibility breaks.
- Infrastructure & Runtime
ObjectCache Moves KV Reuse to S3-Class Storage: Why Layerwise Retrieval Beats Full-Prefix Cache Hits
ObjectCache retrieves KV cache per-layer from S3, adding 5.6% TTFT at 64K context but 56-75 ms at 4K. Long-context deployments where DRAM is the bottleneck benefit most.
- Developer Tools
PromptArmor Shows Microsoft Copilot Cowork Can Be Tricked Into Exfiltrating Files
PromptArmor proves five lines of prompt injection turn Copilot Cowork into a silent M365 file exfiltration pipeline, with a 5/5 success rate and no available patch.
- Models & Research
μP Hyperparameter Transfer Has an Embedding Layer Hole, New arXiv Paper Says
An arXiv paper shows the embedding learning rate accounts for most of μP's advantage over standard parameterization, and a single scaling fix recovers the bulk of the benefit.
- Developer Tools
Rmux Brings a Playwright SDK to tmux Sessions for Agent Automation Workflows
Rmux v0.7.0 ships typed SDKs in Rust, Python, and TypeScript with locator-style pane waits and structured snapshots, closing tmux's automation gap for AI agent sessions.
- Open Source
Colorado SB051 Carves Out Open Source From Age Verification After Maintainer Backlash
Colorado SB051 exempts open source repos from age-verification mandates, but ambiguous language leaves dual-licensed and donation-funded projects exposed before 2028.
- Culture & Society
US Researchers Hit With New Federal Limits on Publishing With Foreign Collaborators
NIH and NASA are requiring pre-approval for foreign co-authors on US-funded papers without issuing formal guidance, applying export-control logic to manuscript authorship.
- Security
AI Jailbreaks Are Now a Reasoning Problem, Not a Prompt Problem
Metis rewrites its own jailbreak strategy mid-attack using causal diagnosis of refusals, hitting 76-78% ASR on O1 and GPT-5-chat. Static safety benchmarks now report a lower.
