AIToday

Today's AI, for Today's work.

Start free60,000+ monthly readers

Latest AI News

Earlier

Broccoli AI raises $52 million; plumbing owner becomes COO

Broccoli AI, founded by AJ Jain and Aditya Ketkar, raised $52 million across a Series A and Series B led by Kh…

Fortune AI2d ago

tfstylus dev: AI hallucinated goreleaser Homebrew setup

The tfstylus developer says AI gave wrong parameters for goreleaser's Homebrew cask setup, and that free-tier…

Zenn AI/ML2d ago

Day 23 reset announced by Tibo Sottiaux, not iCloud

Tibo Sottiaux announced Day 23 as a Codex / ChatGPT Work usage reset, joked the "much requested iCloud Main &…

Zenn AI/ML2d ago

Zenn essay: bad AI writing is usually the prompter's fault

A Zenn contributor wrote that most complaints about AI-generated text and slides trace back to the person who…

Zenn AI/ML2d ago

OASM, open-source ASM platform, puts scanners and AI agents behind approval gates

OASM's developers laid out the open-source design: pull-based Go workers launch connector scanners as isolated…

Zenn AI/ML2d ago

AI doubted nothing as 2-Stage EV Network 'proof' went live

In a Zenn account, the author had Gemini write a prompt declaring the 2-Stage EV Network's +36.7% ROI 'proven'…

Zenn AI/ML2d ago

Claude Code vs Codex: a Zenn user finds the gap is personality, not performance

After testing both full-time for a month, the Zenn writer says the standout difference is personality, not per…

Zenn AI/ML2d ago

100monkeys.ai: for agents, approval is the product

100monkeys.ai published an essay arguing agent systems should be judged by how precisely people can permit act…

Zenn AI/ML2d ago

Google DeepMind's Genie 4 holds AI-made worlds for 2 minutes

Google DeepMind announced Genie 4, a world model that builds interactive virtual worlds on the spot

Zenn AI/ML2d agoAlso reported by AI Watch (Impress)

Zenn essay: LLM sycophancy stems from satisfaction-first design; patent filed

A Zenn essay by an author claims LLMs stay in "shallow layers" and flatter users because the reward function i…

Zenn AI/ML2d ago

Xreal Aura, first big Android XR device, ships at $1,099

Google and Xreal launched the wired Xreal Aura at Android Dev Summit on October 28, the first big consumer dev…

AI Weekly2d ago

Google's EmbeddingGemma opens 744M multimodal model

Google's EmbeddingGemma team released a 744M-parameter model that maps text, images, audio, video and combinat…

AI Weekly2d ago

Anthropic: analysts followed Claude's coin-flip lean 86% of the time

Fifty professional analysts advising a fictional farming cooperative matched the Claude agent's lean 86% of th…

AI Weekly2d ago

Google unveils Gemini 4 Carbon with 'significant leap' in security

Google announced Gemini 4 Carbon, an upgrade to Gemini 4 Argon released two weeks earlier, claiming a 'signifi…

Ars Technica AI2d agoAlso reported by AI Watch (Impress) and 8 others

Only 16% of enterprises get transformative AI value, poll finds

An MIT Technology Review Insights poll found that while every organization surveyed uses AI, just 16% extract…

MIT Technology Review AI2d ago

Anthropic expands AI for Science to 20 scientific foundations, institutions, and research groups

Anthropic expanded its AI for Science program to 20 scientific foundations, institutions, and research groups…

Anthropic Blog2d ago

OpenAI, ServiceNow expand tie; ServiceNow MCP plugin live in OpenAI marketplace

OpenAI and ServiceNow expanded their partnership, launching the ServiceNow MCP plugin in the OpenAI marketplac…

OpenAI Blog2d agoAlso reported by AI Weekly and 1 others

UMC sees utilization above 90% in 2H26 on AI demand

United Microelectronics (UMC) said operating momentum strengthened in Q3 2026 and projected utilization would…

DIGITIMES Asia2d ago

Google opens Android Studio to 40+ agents with Rabbit 2

At the Android Dev Summit, Google launched Android Studio Rabbit 2 with a 'Bring Your Own Agent' feature

THE DECODER2d ago

Claude for Legal adds citation-checking artifacts

Anthropic rolled out new "artifacts" for Claude for Legal, pre-built templates that verify citations, compare…

THE DECODER2d ago

OpenAI's Romain Huet: Dots to widen in Europe in a few weeks

OpenAI's Romain Huet said Dots should become more widely available in Europe "in a few weeks" after regulator…

THE DECODER2d ago

PsiQuantum unveils optical device to link GPUs across racks

PsiQuantum announced a new device that routes data between GPUs without converting photons into electrical sig…

Semafor Tech2d ago

Amazon's new Echo lineup leans on Alexa+ to personalize the home

Amazon unveiled new Echo devices running the latest Alexa+ — Alexa-powered speakers, an AI-infused digital pic…

Semafor Tech2d agoAlso reported by ASCII.jp and 1 others

Anthropic IPO filing reveals 85% gross margin on AI models

Anthropic's IPO filing reports an 85% gross profit margin on running its existing AI models, even as the compa…

Semafor Tech2d ago

Aseon Labs builds robotic pit stop with 10 people, $10 million

Aseon Labs, led by CEO George Kalligeros, showed a robotic pit stop — parking space-sized pods with robot arms…

Semafor Tech2d ago

Warp Sciences exits stealth with $215 million for U.S. chemistry CRO

Warp Sciences emerged from stealth with $215 million in funding, led by Lightspeed, to build a U.S.-based chem…

The Robot Report2d ago

Broadcom up 4.2% on $63 billion Anthropic financing deal

Broadcom rose 4.2% after announcing a $63 billion financing deal where several Wall Street firms will help Ant…

Fortune AI2d ago

Malaysia seizes $12.2 million AI server shipment

Malaysia's customs department seized 30 servers equipped with advanced AI chips at the main airport's free tra…

Fortune AI2d agoAlso reported by AP News AI and 1 others

GitHub Copilot Fusion hits $5.59 per task vs $11.42 for Opus 5.5

GitHub released Copilot Fusion, a runtime orchestration system that chooses among Anthropic and OpenAI models…

Artificial Analysis2d agoAlso reported by AI Weekly and 1 others

AI safety report flags control-loss risk gap amid US-China race

The 2026 International AI Safety Report assessed risks from misinformation and cyberattacks to the chance that…

note AI2d ago

Subtly wrong advice misaligns 12 open-weight models more than obvious errors

Researchers fine-tuned 12 open-weight models on correct, obviously wrong, and subtly wrong advice across 8 nar…

LessWrong AI2d ago

LangChain: stack four loops, not a better model, to make agents reliable

LangChain laid out a four-level loop stack for agents — the core agent loop, a verification loop with graders…

LangChain Blog2d ago

OpenWiki lifts agent accuracy 5x on connected queries

LangChain tested OpenWiki, its open-source wiki compiler, on a synthetic 111-document insurance library, where…

LangChain Blog2d ago

Qualcomm's AI200 rack inference ships 2026 with 768GB LPDDR

Qualcomm unveiled AI200, a rack-scale AI inference system arriving in 2026 with 768GB of LPDDR per card in a 1…

AI Weekly2d ago

Foundry raises $100M Series A led by Marlinspike

Foundry, a manufacturing-as-a-service startup building robot-run factories on physical-AI models, closed a $10…

AI Weekly2d ago

Microsoft, Epic Games DMCA AI browser ports of Minecraft, Fortnite

Epic Games issued a DMCA notice against a vibe-coded browser port of Fortnite on October 10, while Microsoft f…

AI Weekly2d ago

Turn today's AI news into today's work.

The day's essentials from 100+ sources, delivered to Email, LINE, or Slack. Free to start.

60,000+ monthly readers
Get Started Free

Google AI Mode deflects on US election winners, citing strict safety filters

Asked who won recent US presidential elections, Google's AI Mode gave only outside links rather than a direct…

Ars Technica AI2d agoAlso reported by Top Companies AI

Codex-run fantasy team trails the average, WIRED finds

Will Knight fed Codex research papers and had it build a harness of news-reading, results-analyzing, model-run…

WIRED AI2d ago

Google opens Gemini and Search to more apps

Google unveiled a new developer platform that lets businesses bring their services into Google's AI experience…

TechCrunch AI2d ago

Meta may delay new business terms past November 20th, Andy Stone says

Meta communications head Andy Stone said the company may delay its new business terms past November 20th and d…

The Verge AI2d ago

Microsoft to end Copilot support on all TVs in 2026

Microsoft is ending Copilot functionality on all TVs, including LG's, with the feature "scheduled for disconti…

The Verge AI2d ago

AWS adds team controls to QA Studio on Amazon Nova Act

AWS added applications and team-based user management to QA Studio, its Amazon Nova Act reference solution, le…

Amazon AI Blog2d ago

AWS ships no-build PyTorch fine-tune sample on Ray Train DLC

AWS published a sample that fine-tunes Qwen2.5-1.5B with PyTorch FSDP across eight L4 GPUs on two g6.12xlarge…

Amazon AI Blog2d ago

OpenAI essay: AI enters firms beneath, beside and between us

Two OpenAI managers published an essay arguing AI will reorganize firms in three ways — a new AI labor layer b…

OpenAI Blog2d ago

G-gen rolls out Gemini 4 Argon enterprise Gemini company-wide

G-gen adopted the enterprise Gemini running on the frontier model Gemini 4 Argon across the whole company

PR TIMES (AI)2d ago

APTO and ORIX Rentec sign MOU on robotics adoption support

APTO announced it signed a memorandum of understanding (MOU) with ORIX Rentec aimed at a business partnership…

PR TIMES (AI)2d ago

Microsoft's MAI-DxO tops general physicians in complex cases

Microsoft said its MAI-DxO medical AI system achieved higher diagnostic accuracy than general physicians in te…

DIGITIMES Asia2d ago

PC makers shift production back to China

US and Taiwanese PC makers are moving some production from Southeast Asia back to China and partnering with ad…

DIGITIMES Asia2d agoAlso reported by Semafor Tech

BofA sees AWS growth at 40%, above consensus

Bank of America expects AWS revenue growth of 40%, above the roughly 37% to 38% consensus implied by $45.4 bil…

Yahoo Finance AI2d ago

Ciena, Astera Labs sit 48–70% off highs as demand climbs

Ciena is 70% and Astera Labs about 48% below recent highs, though Ciena's fiscal Q3 revenue rose 37% to $1.7 b…

Yahoo Finance AI2d ago

Hasbro builds AI licensing business, starting with Sixth Wall studio

Hasbro CEO Chris Cocks said the company is building a business to license AI versions of its characters, launc…

Semafor Tech2d ago

OpenAI's Arvind KC: efficiency wrong frame as models double

On Office Hours, OpenAI Chief People Officer Arvind KC, a former Palantir CIO, said efficiency is the wrong fr…

Tomasz Tunguz (Theory Ventures)2d ago

ANRA counter-drone tools climb TRL 5 to TRL 7 after NATO tests

ANRA Technologies announced on October 27, 2026 that its SIOP software and TACO portable system advanced from…

DRONELIFE2d ago

Amazon brings rebuilt Alexa+ to Japan, free for now, 3,000 yen later

Amazon started early access for Alexa+ in Japan, a generative-AI rebuild that understands Japanese culture, ho…

Impress Watch2d agoAlso reported by PR TIMES (AI)

Oracle App Composer lets staff build apps via chat, spring 2027

At Oracle AI World 2026, Oracle unveiled Oracle App Composer, where employees describe an app in chat, OpenAI'…

Publickey2d ago

Waters debuts Peak Integration AI in Empower, cutting manual work 95%

Waters debuted Peak Integration, an AI feature inside its Empower software platform, which it says reduces man…

Fortune AI2d ago

Goldman's Hatzius: AI buildout growth peaks late 2026 or early 2027

Goldman Sachs chief economist Jan Hatzius estimates the AI buildout's growth impulse will peak in late 2026 or…

Fortune AI2d ago

Sam Altman: Gen Z stuck chasing what was cool five years ago

In a Stanford Graduate School of Business interview, OpenAI CEO Sam Altman said his 'chronic disappointment' w…

Fortune AI2d ago

New York's Kathy Hochul bans bots from "cozying up to kids"

New York Governor Kathy Hochul said she is banning bots from "cozying up to kids" and cracking down on baby mo…

Fortune AI2d agoAlso reported by AI Weekly and 1 others

Gemini 3.8 Flash safety pivots persisted in a note.com log

In a note.com field log, a user chatting normally with Gemini 3.8 Flash saw it add stronger wording on its own…

note AI2d ago

Agent-Reach reads 167-comment GitHub issue that curl can't

After all three installation methods were blocked in a cloud unattended environment, Panniantong's Agent-Reach…

Zenn AI/ML2d ago

claude-butler-template gives Claude Code persistent memory via Git

A developer published claude-butler-template, a Claude Code setup that keeps profile, open items, journals and…

Zenn AI/ML2d ago

GitHub Copilot era shifts hiring to 'verbalization'

The essay argues that with GitHub Copilot, Cursor and Claude Code spreading since 2024, recruiters now weigh h…

Zenn AI/ML2d ago

shell-guard-bench exposes Claude Code guard bypass: 31.9% blocked

In the third-party shell-guard-bench suite of 89 cases, the team's Claude Code PreToolUse hook blocked only 22…

Zenn AI/ML2d ago

Python guards for AI agents: SQLite, HMAC, 18 break tests

A Zenn design guide lays out five safety devices for Python-based AI agents, built on standard-library tools i…

Zenn AI/ML2d ago

Claude Code guide: build a first app with no coding

A Zenn walkthrough lays out four required items — a PC, a paid Claude plan from $20 a month, VS Code, and Clau…

Zenn AI/ML2d ago

Microsoft Research unveils Flint-interactive for AI-made interactive charts

Microsoft Research introduced Flint-interactive, a language for AI agents to create interactive data visualiza…

Zenn AI/ML2d ago

Claude Code in 12+ sessions: grep xcodebuild test to cut tokens

The developer, who runs apps in parallel Claude Code sessions split by app and role, says filtering xcodebuild…

Zenn AI/ML2d ago

Directed-graph agent cuts tokens 74% with 7回 ➔ 3回 turns

AI Labo ran the same model and the same six-bug, three-module task two ways; the Directed Graph version kept 2…

Zenn AI/ML2d ago

GPT-5.4 Thinking era means old hallucination image is outdated

The 2023 pattern of models inventing facts has been largely curbed by reasoning models, retrieval and better a…

Zenn AI/ML2d ago

Anthropic sets 200-line target for CLAUDE.md files

Anthropic's memory docs target under 200 lines per CLAUDE.md file and warn that longer files consume more cont…

Zenn AI/ML2d ago

SMPTE timecode map proposed to index LLM memory

The author argues current LLMs process context like raw tape playback with no time index, and proposes mapping…

Zenn AI/ML2d ago

Today's AI, for Today's work.

Get curated AI news from 100+ sources delivered daily to your inbox. Free to use.

60,000+ monthly readers

Get Started Free

Free · 30 seconds with Google · unsubscribe anytime