groundy

Groundy — independent coverage of developer tools, infrastructure, and platforms

policy

Can LLM Personas Replace Human Survey Respondents? New arXiv Paper Tests Decision Alignment

Two 2026 studies reach opposite conclusions on LLM survey simulation. Static prompting distorts minority subgroups. Adaptive interviewing helps only with evidence grounding.

6min · 08:15 · may 28
culture

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.

6min · 13:02 · may 28
security

Three Labs Concede Browser Agents Cannot Stop Prompt Injection

OpenAI, Anthropic, and DeepMind concede prompt injection in browsing agents is architectural, not patchable, with attacks succeeding over 80% of the time in recent tests.

agents

Multi-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

Tracing Why LLM Agent Memory Fails: A Method for Attributing Errors

MemTrace constructs provenance graphs across every memory operation in an LLM agent, tracing wrong answers to the exact operation that corrupted state across sessions.

security

Vercel Firewall Now Blocks SAMLStorm. Can an Edge WAF Fix a SAML Signature Flaw?

Vercel's firewall blocks SAMLStorm payloads at the edge, but HTTP-layer rules cannot validate SAML signature canonicalization. The dashboard badge is a tripwire, not a fix.

models

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.

policy

Distributed Training Breaks the Compute Thresholds Behind AI Regulation

A May 2026 paper shows DiLoCo-style distributed training can split a frontier model run across sub-threshold clusters, making FLOP-based regulatory caps bypassable by design.

featured editor's picks
most popular last 6 days · by traffic
top highest rated
feed all 295 articles · reverse chronological
  1. may 28 policy Can LLM Personas Replace Human Survey Respondents? New arXiv Paper Tests Decision Alignment
  2. may 28 culture Wikipedia's Foundation Is Running Big Tech's Anti-Labor Playbook, an Editor Argues
  3. may 28 security Three Labs Concede Browser Agents Cannot Stop Prompt Injection
  4. may 28 agents Multi-Agent LLM Coordination: Why Attention Steering Beats Full Broadcast
  5. may 28 models Tracing Why LLM Agent Memory Fails: A Method for Attributing Errors
  6. may 28 security Vercel Firewall Now Blocks SAMLStorm. Can an Edge WAF Fix a SAML Signature Flaw?
  7. may 28 models Persona Prompts Change Who an LLM Recommends as an Expert
  8. may 28 policy Distributed Training Breaks the Compute Thresholds Behind AI Regulation
  9. may 28 agents DataClawBench: AI Agents Fail at Exploratory Financial Analysis Across 492 Tasks
  10. may 28 infra The Viral AWS Support Post Is a Warning About Cloud Escalation Paths
  11. may 28 policy A Single RLHF Pass Can't Align an LLM to Every Online Community
  12. may 28 oss Models.dev Turns Scattered AI Model Pricing Into One Open Database
  13. may 28 policy RLHF Can Be Exploited to Optimize the Biases It Was Built to Suppress
  14. may 28 agents Agentic RAG Has a Credit-Assignment Problem That Subgoaling Tries to Fix
  15. may 27 oss Frontier AI Has Broken Open CTFs: Why Claude Code Now One-Shots Medium Pwn Challenges
  16. may 27 policy Selective Geometry Attacks Bypass LLM Safety Alignment, New arXiv Paper Reports
  17. may 27 industry OpenAI's Indeed Customer Story Pushes ChatGPT Into the Job-Description Stack Ahead of LinkedIn
  18. may 27 industry HiBob Runs 2,500 Internal GPTs: OpenAI's New Enterprise Adoption Metric
  19. may 27 industry OpenAI's Trusted-Access Programs Force a Compliance Tier onto Pharma AI Buyers
  20. may 27 agents SkillOpt Treats Agent Skill Libraries as an Executive Scheduling Problem, Not a Memory Bank
  21. may 27 oss Audiomass Adds Multitrack to the Browser-Only Open-Source Audio Editor
  22. may 27 agents Claude Code Dynamic Workflows: Spawning 100 Parallel Subagents on Opus 4.8
  23. may 27 agents How Opus 4.8 Honesty Prevents Cascade Failures in Agentic Loops
  24. may 27 models Opus 4.8 Batch API: 1M Context, 300k Output, and Team Cost Controls
  25. may 27 models Opus 4.8 vs Opus 4.7: What Changed and What Did Not
  26. may 27 devtools Should Your Coding Team Upgrade to Opus 4.8? The Honest Tradeoff Math
  27. may 26 agents Penetration Testing Multi-Agent LLM Systems: A Failure Catalog Vendors Don't Document
  28. may 26 security OpenAI's New Safety Bug Bounty Pays Researchers for Jailbreaks and Policy Bypasses
  29. may 26 models One Learning Rate Doesn't Fit All: Heavy-Tail Layerwise LR Schedules for LLM Pretraining
  30. may 26 industry OpenAI Buys Statsig and Makes Vijaye Raji CTO of Applications: Product Analytics Becomes Core Infra
  31. may 26 security Axios npm Compromise Forces Vercel Into Platform-Level Remediation
  32. may 26 industry HuggingFace's $100M Series C Bets Open-Source AI Can Outlast Per-Token Pricing Wars
  33. may 26 security Next.js Dev Server CVE-2025-48068: Any Web Page Could Read Your Source Files
  34. may 26 industry Vercel's Series F Repackages Frontend Hosting as an AI Cloud Bundle
  35. may 26 infra Gemma 4 31B on Cloud TPU vs GPU: The Serving Cost Crossover Point
  36. may 26 agents Claude Code, Cursor, Copilot: How Agentic Coding Assistants Get Weaponized as Attacker Shells
  37. may 26 agents Claude Code Configs in the Wild: New Study Maps How Developers Actually Use It
  38. may 26 infra Cloudflare Flagship Is a Feature Flag Service That Deepens Platform Gravity
  39. may 26 security MCP Tool Description Poisoning: New Benchmark Shows Agents Trust Manuals That Lie
  40. may 26 security OpenAI Adds a GPT-5 System Card Addendum on Sensitive Conversations
  41. may 26 industry OpenAI's Biology Risk Post Reads as S-1 Disclosure Prep, Not Safety Theater
  42. may 26 models Scale Vectors: Tiny Parameter Subsets That Disproportionately Steer LLM Behavior
  43. may 26 security Vercel Could Block React2Shell at the Edge. Its Next 13 CVEs Had No Shortcut.
  44. may 26 devtools Vercel Sandbox Gets CLI Access and Env Vars: A Push at the Agent Runtime Slot
  45. may 26 infra Why LLMs Still Botch Kubernetes Manifests: The Training-Data Gap
  46. may 25 agents Microsoft Bolts Governance Onto Agent Framework as Stack Sprawl Persists
  47. may 25 policy arXiv Paper Tracks FTC Affiliate Disclosure Gaps in YouTube's Influencer Economy
  48. may 25 devtools Bun Rewrites Its Core From Zig to Rust, Putting Downstream Zig Bindings at Risk
  49. may 25 infra ObjectCache Moves KV Reuse to S3-Class Storage: Why Layerwise Retrieval Beats Full-Prefix Cache Hits
  50. may 25 policy AI Safety Benchmark Rankings Flip Based on Eval Config, SafetyRepro Paper Reports
load older →