news

AI Tech News

Submit

Artificial intelligence remains the fastest-moving story in technology. Here you will find news on model releases and research, the labs and companies competing for the frontier, enterprise adoption, and the debates over safety, jobs, and regulation that AI continues to raise.

Typical stories include new model launches and benchmark claims, massive infrastructure and chip investments, AI features arriving in mainstream products, research papers that shift the field's direction, and enterprise deals that show where adoption is real rather than experimental.

AI now sets the agenda for the rest of the technology industry, steering capital, talent, and strategy at nearly every major company. Developers, executives, policymakers, and workers wondering how their jobs will change all have reasons to follow it closely.

2026-11-12

2250

Nvidia CEO Jensen Huang has resigned, capping a spectacular downfall for a poster boy of the artificial intelligence boom. The resignation marks a major leadership shake-up at the chipmaker.

Comments
329

Google has introduced Gemini 4 Hydrogen, a fast and cost-efficient model designed for reliable agentic workflows and everyday tasks. It is positioned for dependable performance across routine and agent-driven use cases.

Comments
258
Anthropic IPO
2026-11-12

Anthropic has filed to become a public company, setting up an IPO for the AI developer. The company directed readers to a founders’ letter and other relevant information accompanying the filing.

Comments
212

When we launched the Remote Labor Index a year ago, the best AI could fully automate 2.5% of real freelance projects. Fable 5.2 now automates 41.3%. The automation rate is doubling every ~3 months. AIs may soon be able to do all remote work faster and more cheaply than humans.

Comments
138

Jensen Huang has resigned as Nvidia CEO immediately, a move readers see as removing the leader who defined the company for over 30 years. Discussion focused on whether Nvidia will be revalued without him and on risks from fines, export controls, China revenue exposure, and AI-bubble sentiment, with some still viewing the underlying chip business as fundamentally strong.

Comments
104

Google is rolling out Gemini 4 Hydrogen inside Google Antigravity. The update also adds an Orchestrator option to the model dropdown, pairing Hydrogen with additional capabilities.

Comments
85

A forensic analysis of publicly released AI models found evidence that DeepSeek, Moonshot AI, MiniMax and Ant Group have gained access to NVIDIA’s export-controlled Blackwell GPUs and used them to build their models. The findings suggest Chinese frontier AI developers obtained the restricted chips despite export controls.

Comments
84

The Foundation Models framework offers a native Swift API for integrating any large language model into an app. It gives Swift developers a standardized way to add LLM capabilities directly into their applications.

Comments
80

Jane Street has added a Socratic Tutor skill to its agent_skills GitHub repository. The repo is described as a collection of agent skills and is open for community contributions.

Comments
70

A developer is working on smolmaxxing, a project aimed at maximizing the ability to run different LLMs on the smallest and oldest devices possible. The effort focuses on making local AI inference feasible on highly constrained hardware.

Comments
67

Libreagi is a private BitTorrent tracker where members keep copies of open-source AI models that are most likely to disappear. The project aims to preserve access to at-risk OSS models through distributed file sharing.

Comments
52

The surge of AI and AI-generated apps is overwhelming Apple’s App Store review process, creating major headaches for developers. Legitimate synth app makers are getting caught in the crossfire as reviews bog down, with the report citing two examples.

Comments
48

A tinkerer automated a two-square-metre vegetable garden with eleven sensors, a camera, an old phone and an AI agent backed by a lakehouse, producing a lettuce that cost about €60. The stunt illustrates how IoT is shifting toward agentic automation and integrated data platforms, while showing the cost and complexity still involved in real-world deployments.

Comments
48

Fable52.games offers 30+ browser games built autonomously by Claude Fable 5.2, many shipped as single self-contained HTML files. Visitors can play the games, read the exact prompts, inspect the source, and download or remix them.

Comments
42

AgentRun is a new open-source platform that adds self-improvement, durable execution, and a control plane on top of existing agent harnesses. It aims to make agents smarter and more fault-tolerant while using fewer tokens with each run.

Comments
28

The article argues that an AI crash should be expected, even as warnings about it become commonplace. Reader discussion highlights the whiplash between fears that the bubble is already priced in and social platforms insisting AI is the new economy that cannot flop, with that constant swing between hype and alarm itself suggesting fragile sentiment.

Comments
21
Lean Grimoire
2026-11-12

The author recounts experimenting since July with modern coding agents to build mathematical visualization applets, a project that had to be set aside during a hectic few months. The piece reflects on that agent-assisted development work and what it suggests for building small interactive math tools.

Comments
19

Kapso is an iterative-learning framework for autonomous program synthesis and optimization on enterprise data, though its name clashes with an existing WhatsApp integration service and the team may rename it. Early users report it efficiently calibrated RelFlow on 21 structured data tasks and beat TabPFN on many larger modeling tasks with ~1% of its parameters, while also claiming strong RelBench results. Commenters want controlled RelArena comparisons and debate whether such experimental learnings should compound in an agent harness or be baked into model weights via RL.

Comments
16

A new Show HN project, AI Red Lines, asks people to define in advance what AI developments would prompt them to take action and what that action would be. The premise is to set personal thresholds while thinking clearly, then share them publicly.

Comments
5

Darling has published its Q4 2026 development update, introducing a new policy on AI-assisted contributions and adding initial ARM64 support. The project also outlined its next work priorities after a long gap since the last progress report.

Comments
4

Cerulion, a YC S24 company, has released an open-source ROS 2-compatible runtime and harness that lets AI agents control real robots within safety limits enforced by the robot. It also gives agents tools to autonomously debug recorded robot failures using multi-process replay.

Comments

2026-11-11

903

AI agents analyzed public lidar data overnight and identified a Maya site beneath the jungle. Archaeologists had already mapped the site, but it had not yet been added to Mexico’s official register.

Comments
339

Google SWE used 48 AI agents to analyze NASA lidar data overnight, uncovering an unmapped Maya site with a pyramid hidden beneath the jungle. Mexico’s INAH responded within 72 hours.

Comments
243

United Airlines has banned AI agents from booking flights on behalf of travelers, citing security risks and possible violations of federal rules. The policy targets automated tools that make reservations for passengers.

Comments
109

A group of Emacs users and developers is pushing for an LLM-free version of Emacs, vowing to continue it even if GNU adopts a different policy and urging GNU to reject LLM-generated contributions. The debate around the effort highlights broader concerns about AI dependency, software freedom, and whether coding agents make programming more accessible or undermine understanding.

Comments
82

MIT CSAIL’s recent “Sparks of Weight-Space Programming” paper showed that GPT-6 Astra could identify which game a small network had been trained on directly from its raw weights. The new piece asks whether a linear program can perform comparable weight-space programming, probing whether such capabilities require large models or can be replicated with simpler linear methods.

Comments
80

A UC Irvine software engineering seminar on October 30, 2026 will examine AI, activity, and outcomes in software engineering. Video, slides, and a script from the talk are available.

Comments
78

AWS Emulator, launched today and written and tested entirely by agents, provides a local copy of more than 200 AWS services for testing. Swami Sivasubramanian recounts an early high-fidelity simulation effort and explores how engineers’ work changes when agents write the code.

Comments
70

Americans overwhelmingly oppose the AI data center buildout, but their retirement accounts are helping fund it anyway. The piece examines how retirement savings are channeling money into the AI boom despite public opposition.

Comments
59

Google has introduced the Gemini Decisions API, offering fast, calibrated multimodal decisions for Google apps. The API can evaluate 1,000 questions in a single pass, supporting high-volume decision-making tasks.

Comments
2026-11-12