HN rajsuper123 19d ago

Claude ends the chat – refuses conversation

Anecdote about Claude AI refusing to continue conversation due to user insults. Opinion piece, limited technical value.

LB blog.dshr.org via fazalmajid 19d ago

Chatbots vs Ozone

Blog post about digital preservation work. Off-topic.

HN anax32 19d ago

Ask HN: How is GPU power draw measured at scale?

Question about measuring GPU power consumption at scale for inference systems. Infrastructure concern, relevant to ML deployment.

HN adu_onemore 19d ago

Analysed 3k+ AI thought leaders posts for Q2

Analysis of AI thought leaders' posts in Q2 covering Anthropic/OpenAI valuations and model performance.

HN yesbut 19d ago

Why AI in the classroom is a catastrophe – Sophie Winkleman [ARC 2026] [video]

Stub article title only, no content provided.

HN FergusArgyll 19d ago

Trump administration asks OpenAI to limit next model release

Stub article title only, no content provided.

HN croes 19d ago

DuckDuckGo's AI Feature Is Telling Users That Trump Died of Rabies

DuckDuckGo AI search feature generates false information about Trump. Criticism of LLM hallucinations, not substantive.

HN vantareed 19d ago

Windows-Copilot-API; Access GPT-4 and GPT-5 models without API keys or billing

Unofficial API wrapper accessing GPT-4/GPT-5 via Microsoft Copilot web interface without API keys. Violates ToS.

BL 19d ago

Previewing GPT-5.6 Sol: a next-generation model

OpenAI announces GPT-5.6 series with three models: Sol (flagship), Terra (balanced, 2x cheaper), Luna (fast, affordable). Includes safety improvements.

HN vincent_s 19d ago

The AI Productivity Trap Is More Output

Stub article title only, no content provided.

HN PennRobotics 19d ago

Show HN: Unofficial cross-platform Shure MV client

Unofficial cross-platform client for Shure microphone adapter, unrelated to AI/ML.

HN raketenkater 19d ago

Llama.cpp flags auto-tuning tool

llama.cpp's ggrun auto-tuning tool for GGUF model optimization across multiple GPUs without manual flags.

HN leporis14 19d ago

Agentlint – A security scanner for MCP server configs

Security scanner for MCP server configurations used by AI agents. Detects dangerous permissions, hardcoded secrets, missing guardrails.

HN ankitg12 19d ago

AI Search Engine Exa Raises $250M Series C

Exa raises $250M Series C at $2.2B valuation to power web search for AI agents and LLM applications.

HN zingar 19d ago

It took two weeks to make Claude's "overnight solution" for flaky tests useful

Engineering case study: Claude AI helped fix flaky tests but required significant iteration to become production-ready.

HN ML0037 19d ago

Show HN: I built a small audit layer for LLM-as-judge decisions

Open source audit layer for LLM-as-judge systems that validates verdicts against supporting evidence.

HN thm 19d ago

California State Government Launches AI Job Loss Tracker as Layoff Fears Grow

California tracks AI-related job losses as layoff concerns grow.

HN merksittich 19d ago

AI 2027 Tracker

Tracks AI capability predictions against actual progress; mentions Agent systems and research acceleration.

HN jeroenjanssens 19d ago

Translating Pandas to Polars using LLMs

LLMs enable rapid Pandas-to-Polars migration for data engineers; increasingly used for code translation tasks.

HN willy__ 19d ago

Paying for LLM inference by the kilowatt-hour instead of per token

Analysis of LLM inference cost models: kilowatt-hour billing vs token-based pricing for open-weight models.

HN tilt 19d ago

Google Vids: AI-Powered Video Creator and Editor

Google's LLM-powered video creation tool using Gemini for storyboarding and media generation.

HN GustavHartz 19d ago

Building effective pen-testing agents

Technical guide on building autonomous agents for penetration testing and security assessments.

HN chandler93 19d ago

CasaOS: An open-source home server OS for Docker apps

Open-source home server OS for Docker apps with personal cloud aspirations. Limited technical depth in excerpt.

HN mariuz 19d ago

Evaluating performance and efficiency of the GitHub Copilot agentic harness

Performance evaluation of GitHub Copilot's agentic capabilities and efficiency metrics.

HN xgboosted 19d ago

Pandas-ta-classic 0.6.52: SMC liquidity sweep detector, Ichimoku fixes

Financial analysis library with technical indicators and candlestick patterns; not AI-focused.

HN adityapatadia 19d ago

Why current LLM costs are not sustainable

Analysis of LLM operational cost sustainability challenges.

HN xdotli 19d ago

A curated, non-BS library of the best resources for evaluating agents

Curated library of resources for building and evaluating AI agents including papers, tools, courses, and benchmarks. Maintained collection with verified annotations and dead links removed.

HN teleforce 19d ago

PatentScore: Multi-Dimensional Evaluation of LLM-Generated Patent Claims

PatentScore introduces a multi-dimensional evaluation framework for assessing LLM-generated patent claims. Published at EMNLP 2025, presents research on evaluating LLM output quality.

HN joozio 19d ago

Notion killing Skiff-influenced email app since most users use AI agents instead

Notion acquired Skiff email and is pivoting to AI agents for inbox management, shutting down Notion Mail. News about enterprise AI agent adoption.

HN VaderAi 19d ago

Ask HN: Why does every AI demo sound perfect but real world deployment always

News article about researcher receiving death threats while computing Rosetta Stone translation for Indus script.

HN novaesystems 19d ago

I made a Claude Code skill to check if AI crawlers can read your site

HN discussion on gap between AI demo performance and real-world deployment challenges in AI voice applications for small businesses.

HN xSingh16 19d ago

Ask HN: How are you solving long-term memory for production AI agents in 2026?

Fivy is an AI-powered review response tool for small businesses that drafts replies to Google reviews and identifies improvement areas.

HN angelmm 19d ago

Local and GitHub Code Review TUI

tuicr is a terminal UI for code review with vim keybindings, GitHub-style diffs, and line-level comments. Integrates with GitHub PRs and supports structured markdown output for coding agents.

HN theanonymousone 19d ago

US Govt to individually approve who gets GPT 5.6

swSIM is an all-software SIM card simulator without relying on hardware, first publicly available implementation.

HN JerBouma 19d ago

Show HN: Every site gives a different P/E ratio, so I open-sourced the math

Reddit post claiming US government will individually approve who gets GPT 5.6. Low-quality speculative content.

HN korbonits 19d ago

An LLM verifier rated math proofs near-perfect; an expert found 17% correct

Research on converting brain prediction models into testable explanations. LLMs predict human cortical activity from language but lack interpretability.

HN tosh 19d ago

Optimus Cirrus

Developer open-sourced P/E ratio calculation code to standardize earnings calculations across financial data providers.

HN bickov 19d ago

I feed my coding agent JSON instead of screenshots

Developer shares SlimSnap tool that feeds JSON to coding agents instead of screenshots to reduce token costs and improve accuracy.

HN tanelpoder 19d ago

Modern GPU Programming for MLSys Book

Educational resource on GPU programming for ML systems, covering attention kernels, LLM kernels, and optimization techniques.

HN RihabAI 19d ago

Context loss is the real reason AI coding slows down engineering teams

Product pitch for Brunelly, an AI software development platform for managing engineering workflows. Promotional content.

HN slow_typist 19d ago

AI coding will be more expensive than human developers

Gartner analysis predicting AI coding costs will exceed developer salaries by 2028 due to token consumption and billing models.

HN wahnfrieden 19d ago

Trump administration asks OpenAI to stagger release of new model

News report on Trump administration requesting AI model review from Meta. Political/regulatory news, not technical.

HN susros 19d ago

Show HN: Appaca – AI Workspace for Operators

HN show: Appaca, an AI workspace for operators using LLMs to generate code. Third iteration of no-code/LLM idea.

HN monoids 19d ago

Ludwig Spec Driven Development MCP

Astryx is an open-source design system built on React and StyleX, originally from Meta's internal tools. Explicitly agent-ready and customizable.

HN teleforce 19d ago

Researcher got a death threat computing Rosetta Stone for Indus script [video]

Ludwig is a specification-driven development framework where prose-first markdown specs drive LLM code generation and verification. Targets Rust with cargo-based deterministic checks.

Ax Xi Xiao, Chen Liu, Chih-Ting Liao, Yunbei Zhang, Qizhen Lan, Yuxiang Wei, Lin Zhao, Janet Wang, Jianyang Gu, Muchao Ye, Tianyang Wang, Hao Xu 19d ago

Staying VIGILant: Mitigating Visual Laziness via Counterfactual Visual Alignment in MLLMs

Method addressing visual hallucinations in MLLMs by mitigating visual laziness through counterfactual alignment.

Ax Luoning Zhang, Xu Zhuang, Tianhao Wang, Nathan Kaplan 19d ago

Geometry-Aware MCTS for Extremal Problems in Combinatorial Geometry

MCTS approach with geometry awareness for solving extremal combinatorial geometry problems efficiently.

Ax Jobst Heitzig 19d ago

Unbiased Canonical Set-Valued Oracles Via Lattice Theory

Theoretical framework for oracle AI systems addressing self-reference problem in probability estimation.

Ax Neelam Saini, Sourav Ghosh 19d ago

Listening Like a Judge: A Music-Aware Framework for Automatic Singing Performance Evaluation

Music-aware framework for automatic singing quality assessment combining acoustic and lyric evaluation.

Ax Xiao Wang, Xufeng Lou, Zikang Yan, Lan Chen, Sibao Chen, Yaowei Wang, Yonghong Tian, Jin Tang 19d ago

Active Adversarial Perturbation-driven Associative Memory Retrieval for RGB-Event Visual Object Tracking

RGB-Event fusion tracking system using adversarial perturbations for robust visual object tracking.