Markbhall
My local LLM scored 6/6. It was wrong every time
The author discovered that a 1.2 billion‑parameter local LLM appeared to pass a six‑question benchmark despite…
ai · source
Hacker News (AI / LLM) on WeSearch
Recent ai headlines from Hacker News (AI / LLM).
Dinjure
Show HN: Dinjure – Bulls and Cows in one HTML file, with a fallible AI
Crack the daily code in 8 guesses — pure logic, no math. Dead = right digit, right spot. Injured = right digit, wrong…
Hacker News (AI / LLM)
Using AI for Automated Tasks
Google DeepMind
Securing the Future of AI Agents
June 18, 2026 Responsibility & SafetySecuring the future of AI agentsRohin Shah and Four Flynn Share CopiedHow we’re…
Zenodo
Tenir: A Counter-Pressure Architecture for Safe Agentic AI Under Irreversibly
Published July 9, 2026 | Version 5.0.0 Software Open TENIR-Gov: Governance Middleware for AI-Enabled Operational…
GitHub
OptMem – Permanent memory for AI agents
Install curl -fsSL https://raw.githubusercontent.com/VictorTaelin/OptMem/main/install.sh | sh It prints a ## Memory…
https://galjot.si/
What AI developers could learn from Charles Bukowski?
What AI developers could learn from Charles Bukowski? I’m sure it will. they might even remember me. – an excerpt from…
ZK/SEC Quarterly
AI meets Cryptography 3: What AI Found in Bron Labs's bron-crypto
AI meets Cryptography 3: What AI Found in Bron Labs's bron-crypto Stefanos Chaliasos, Hao Pham, False Witness Team…
Github
XMPP Agent Gateway: A protocol for discovering and invoking AI agents
XEP-xxxx: XMPP Agent GatewayAbstractThis specification defines an XMPP protocol extension for hosting, discovering,…
DynoTable
Show HN: DynoTable – The DynamoDB GUI with real SQL and your AI agent
The DynamoDB GUI with real SQL & your AI agentA fast, local-first desktop client for DynamoDB. No third-party service…
MIT Technology Review
Building the enterprise environment for agentic AI
SponsoredArtificial intelligenceBuilding the enterprise environment for agentic AIEnterprises will find success with a…
GitHub
Show HN: Preloop: Open-source control plane for AI agents. See, govern, cut cost
Preloop is an open-source platform that provides a control plane for AI agents, integrating tool access firewalls,…
GitHub
Show HN: Banksia – Build and run adaptable, accountable AI teams
Banksia is a Python‑based platform that lets users build and manage adaptable AI teams through a visual console or…
Puck - Create your own AI page builder
Puck AI – Visual AI builder embedded in your app
AI has transformed how software is built, but most of the productivity gains have gone to developers. Meet Puck AI…
Hacker News (AI / LLM)
AI Latency Is a Feature, Not a Bug
AI Latency Is a Feature, Not a BugFilling the AI Latency gap with more agents is the new Mythical Agent-MonthMarc…
P. Galeone's blog
Use Your Brain: Engineering Standards in the Age of LLM
The advent of LLMs made it possible for everyone to write their own software - and this is a good thing. However, I…
Segue
Show HN: Segue – Save context in one AI, load it in another by a short handle
Works with Claude · ChatGPT · any MCP client Hand context fromone AI to the next. Save a block of context in one…
Hacker News (AI / LLM)
AI revenues are growing fast, but not fast enough
GitHub
Show HN: Hotcell – local sandboxes for AI agents
hotcell Sandboxes for AI agents, on your own hardware. A Mac Mini on your desk, a cloud VM, a bare-metal box — as many…
GitHub
Microsoft Quicksand: Sandbox your AI agent without Docker or WSL
Quicksand Quicksand is an async Python API to launch, control, and snapshot QEMU virtual machines with a particular…
Antirez
The AI risk is inside the labs
Bytebytego
How Nvidia Builds Open Models for the Age of AI
How NVIDIA Builds Open Models for the Age of AIByteByteGoJul 27, 2026236116ShareChainable compute. Right on queue.…
Collabora Online and Collabora Office
Collabora Online 26.04: Your Documents, Your AI, Your Decade of Data Sovereignty
Collabora Productivity delivers Collabora Office, a business-hardened, open-source office suite providing seamless…
Business Insider
Amazon overhauls its AI strategy, winding down most flagship models
Amazon reshapes its AI strategy by deprecating most Nova AI models and shifting resources to Pieter Abbeel's Frontier…
Hacker News (AI / LLM)
AI Coding is Exhausting [video]
Rosebud AI
The Prompt Atlas: a map of the games people ask an AI to build
Thousands of real, unedited game ideas typed into Rosebud AI, mapped into one galaxy. Every star is one real request.
STAR Labs
When AI Makes 0-Days Feel Like N-Days
The author used AI to locate a use‑after‑free vulnerability in the Linux kernel’s net/sched subsystem and crafted a…
arXiv.org
LoopLynx: A Scalable Dataflow Architecture for Efficient LLM Inference
LoopLynx is a scalable dataflow architecture designed to improve large language model (LLM) inference on FPGA…
Hacker News (AI / LLM)
Google's Beyond Zero: Enterprise Security for the AI Era
Zenodo
How AI Can Locate Hidden Defence Labs and Military Installations Without Breach
Published July 27, 2026 | Version v1 Publication Open How AI Can Locate Hidden Defence Labs and Military Installations…
Ycombinator
Ask HN: Thoughts on AMD Ryzen AI MAX+ 395 for Local AI?
Own a framework desktop 128gb.GPU bandwidth limits the amount of tokens/sec that you get.I 've mostly been running…
Discounthub
Show HN: DiscountHub – AI-assisted search for deals and promo codes
DiscountHub is a web-based platform that uses AI to help users find real discounts, promo codes, and store offers. It…
Ycombinator
Ask HN: Is there legal risk in AI memory?
The post raises concerns that AI systems that record and embed meeting content could create legal liability during…
Tnt-audit
Risk-Data API – Solana token risk scoring for AI trading agents
For Claude, Gemini, or agent frameworks like LangChain/CrewAI, use this spec as the schema source for your own tool…
RenderPop AI
Show HN: RenderPop – Free browser-based AI image and video generator
{"@context":"https://schema.org","@type":"SoftwareApplication","name":"RenderPop","url":"https://www.renderpop.app","ap…
Antirez
The AI risk is inside the labs
Clarity
Soumitra Dutta: A Visionary Perspective on AI and Global Competitiveness
Soumitra Dutta is one of the very Indian academician as well as an entrepreneur who is famous for co-inventing an…
GitHub
Show HN: BrowserAct: Browser Layer for Your AI Agent
BrowserAct Skills Browser automation CLI built for AI agents. Get past anti-bot walls, hand off to humans across…
WizeMe.APP
Show HN: WizeMe – governed AI continuity with user-controlled memory
WizeMe is a user‑controlled AI continuity platform that links personal and work data while keeping memory…
Mergify
AI writes code that breaks main half as often
On the benchmark cohort, a green PR breaks main about 0.77% of the time at 2 to 5 engineers (1 in 130), but 12.5% of…
nextplatform
AI Hosts and Sandboxes Save Intel's Datacenter CPU Cookies
Yes, we all thought that the CPUs used in AI host servers would have lots of strong cores for AI hosts and need lots…
GitHub
Show HN: OpsCat – A single-binary software catalog with MCP for AI agents
🐾 OpsCat The Zero-Config, Single-Binary Software Catalog & MCP Platform Engine Auto-discover microservices, visualize…
Probably Dance
If AI Writes All the Code, What Do the Programmers Do?
Eight months ago I was producing roughly 90% human written code and 10% AI written code. This has switched…
Ycombinator
Tell HN: Our paid Claude AI subscription unavailable >1 week and no support
A user reports being incorrectly billed after canceling a Claude AI subscription and receiving only automated bot…
Bbc
Some people's chats with Claude AI found to be publicly available online
Anthropic's Claude chatbot conversations were inadvertently exposed via search engines when users shared links, making…
VideoLens
Show HN: Reverse-engineer viral videos into AI-ready shot breakdowns
{"@context":"https://schema.org","@type":"WebApplication","name":"VideoLens","url":"https://videolens.cc/zh/landing","i…
Metadata Killer
Metadata Killer – Remove EXIF, GPS, AI Metadata, and C2PA from Images
The Metadata Killer is an online tool that removes EXIF, GPS, AI metadata, and C2PA information from image files. It…
The Conversation
Cracking the code: can AI help us decipher ancient languages?
Ancient languages such as Linear A and Etruscan lack the bilingual anchors that have traditionally enabled…
nilenso blog
Evals before prompts: building an LLM OCR for KYC
The article describes how the author built an LLM-based OCR system to extract data from Indian KYC documents such as…
AP News
AI 'tokenmaxxing' fades as workplaces look to cut tech spending
The practice of “tokenmaxxing,” or maximizing AI token usage, is losing popularity as companies curb tech spending.…
How WeSearch handles this source
WeSearch's declared handling of Hacker News (AI / LLM)'s content. Indexing, snippets, summaries, retrieval and training are separate questions — see the rights registry or read this source's machine-readable record.
Indexing: Allowed
Snippet: Allowed
AI summary: Limited
Retrieval / RAG: Not asserted
Model training: Not asserted
Commercial reuse: Not permitted