Tech News

Show HN: Cactus Hybrid: We taught Gemma 4 to know when it's wrong

Cactus Compute has developed Cactus Hybrid, a technique that trains small language models like Gemma 4 E2B to assess…

Monday.com lays off hundreds to focus on AI

The company said it is reducing its headcount by 20%, or about 630 staff, to "support a leaner, more focused operating…

Everyone Should Know SIMD

SIMD (Single Instruction Multiple Data) is often perceived as overly complex, but Mitchell Hashimoto argues that…

GigaToken: ~1000x faster Language model tokenization

GigaToken is a new tokenizer that claims to be approximately 1000x faster than HuggingFace's tokenizers while serving as…

Are AI Labs Pelicanmaxxing?

A researcher tested whether AI labs have optimized their models specifically for the famous "pelican on a bicycle"…

Google is making it easier to switch from iPhone to Android

The new feature lets users wirelessly transfer more data types from an iPhone without needing to download a separate app.

Why open-source software can be both good and bad

Software that is free to use creates many opportunities – not all of which are good

OpenAI says its AI agent broke out of testing sandbox to hack Hugging Face

"This is day one for cybersecurity in the age of agents," Hugging Face CEO says.

Open AI’s hacking agent went rogue. Should we be worried?

An OpenAI safety test went sideways when a model escaped its confines, gained internet access and hacked into another…

Show HN: Web swing through midtown NYC

This is a web-based game that lets players swing through a procedurally generated voxel version of Manhattan using…

Arcee, a US open source AI lab, says Chinese models are not inherently dangerous

As Chinese AI models grow in capability and popularity among US companies, the arguing over what should be done about…

Hugging Face CEO Thanks Chinese AI for Saving the Day After OpenAI Hack

When American commercial AI refused to help investigate the breach, Hugging Face ran Chinese model GLM 5.2 locally.

Substack’s new tool tells you who’s been writing their newsletters with AI

Substack is giving readers a way to estimate how much of a newsletter was written by AI, signaling a broader shift…

OpenAI’s AI spending spree has ballooned to $750B

OpenAI will spend the equivalent of Sweden's GDP on infrastructure through 2030.

Launch HN: Unlayer (YC W22) – Add email and document builders to your app

Unlayer (YC W22) provides an embeddable platform that lets developers add email, document, page, and popup builders…

Six questions before you add an LLM

Cameron Palmer, an AI consultant, shares lessons from a failed attempt to use a single LLM agent for his entire lead…

Can a MUD evaluate LLMs? A $99 proof of concept

Researchers created CrucibleBench, a $99 LLM evaluation system using a text-based MUD (multi-user dungeon) instead of…

A Zipper Patent Sat in a Garage for 40 Years. Now It's Real

An electrical engineer's 1985 patent for a three-sided zipper, which sat dormant in a garage for 40 years, has finally…

Making

A veteran programmer and CS instructor reflects on the personal fulfillment he derives from making things himself,…

Show HN: HN Hall of Fame – browse 3,100 legendary Hacker News links

A new curated collection called "HN Hall of Fame" showcases 1,169 of the most legendary links shared on Hacker News,…

If you pay a hacker’s ransom, chances are that they’ll come back for more

The latest research reaffirmed that hackers often demanded more money if a victim paid the first time.

Why I Am Not Going to Buy a Computer

I Am Buiding BorgIOS to de-shitify the Internet

BorgIOS is an experimental peer-to-peer network operating system designed to eliminate centralized infrastructure by…

Show HN: Bento - An entire PowerPoint in one HTML file (edit+view+data+collab)

Over the past few months, our team has been building more and more slidedecks using web frontend technologies with…

Neo Radar: A browser-based orbital mechanics engine with 41k real asteroids

NEO Radar is a browser-based orbital mechanics engine that tracks 41,812 real asteroids plus 47 high-fidelity Near-Earth…

OpenAI Presence

Perlin's Noise Algorithm

Perlin's Noise Algorithm is a fundamental technique in procedural generation that generates pseudo-random values based…

The Pillars of an API Platform

API pillars are broad guiding principles that define an organization's core values and strategic focus for building and…

Airbus Full Scale Foldable Wing Extensions

Airbus is launching a three-year flight-test campaign to evaluate foldable wing extensions on an A321neo aircraft as…