Latest AI News

Get the fastest, most accurate AI news every morning

Total articles: 12,318|Added todayToday: 139

South Korea Restructures Frontier AI Project

Government separates Frontier AI from its foundation-model program, shifting to a separate open call and investment structure.
Officials had considered assigning development to the two companies selected in the foundation-model program’s third evaluation.
The government is considering raising GPU support from 1,000 to more than 10,000, prompting concerns about concentration at one company.

ZDNet KoreaRead more · 2 min →

Today'sInfra

KIND Backs Jakarta AI Data Center Project

KIND invested in a Jakarta AI data center, supporting Korean companies’ first entry into the overseas commercial data center market.
Korean firms secured about $107 million in overseas orders through construction, operations and equipment supply.
LG CNS secured contracts worth at least $70 million; KIND-linked investment orders totaled $37 million.

NewspimRead more · 2 min →

Today'sSafety & Ethics

AI Glasses’ Recording Blind Spot Exposed at Audit

A demonstration at a parliamentary audit showed that attendees could not identify who was recording with AI glasses.
Police confirmed three suspected illegal-filming cases involving smart glasses this year, including two on aircraft.
The science ministry plans cross-agency measures and usage guidelines amid concerns over exam cheating and covert filming.

Dong-A IlboRead more · 2 min →

Today'sInfra

South Korea Eases AIDC Approval Rules

Government announces simpler AIDC permits and measures to support the domestic ecosystem
Next year’s proposed budget allocates about 200 billion won for local technology development and testing
AIDC special zones planned for the first half of next year will require community benefit plans

Financial NewsRead more · 2 min →

Today'sHealthcare

AI Scribes Save Time but Leave Burnout Unresolved

More than 12,000 Canadian primary care clinicians enrolled in a national AI scribe program; almost 60% report less administrative time
In tests of 20 approved Ontario AI scribes, 12 recorded wrong drugs, nine invented treatment steps, and 17 missed mental health details
Authors propose GRACE principles and an adoption matrix, urging health systems to track burnout, safety, and whether saved time returns to clinicians

medicalxpress.comRead more · 2 min →

Today'sInfra

Businesses Face Internal Barriers to AI Transformation

Microsoft engineers worked with sales staff to reshape Copilot tools after employees stopped using them
Cognizant survey: 32 percent reported productivity gains; 60 percent greater pause risk with nontechnical investment
Executives cited infrastructure, oversight, regulation, and employee support as barriers to enterprise AI adoption

newsweek.comRead more · 2 min →

Today'sSecurity

PHP Vulnerabilities Boost OpenAI’s Cybersecurity Push

Research teams found two AI-assisted PHP vulnerabilities in August; the flaws affect software used by an estimated 70% of the web.
Amazon Web Services said hackers began exploiting php2shell on Oct. 17, before the public disclosure and patch.
OpenAI says Daybreak participants found 1,885 software flaws and produced 661 patches, as the company builds cybersecurity sales.

FortuneRead more · 2 min →

Trending Keywords

Today'sSafety & Ethics

Google DeepMind Employees Resign Over Pentagon AI Deal

Google DeepMind employees resign over a Pentagon contract allowing AI use in classified settings
Sofia Liguori left after three years; Andreas Kirsch cited insufficient oversight of military AI
More than 600 employees urged Sundar Pichai to block classified Pentagon use of Google AI

breitbart.comRead more · 2 min →

Today'sSafety & Ethics

Altman Says Mathematicians Are Embracing AI

Sam Altman says mathematicians are moving from fears of field-wide disruption toward different career paths
OpenAI’s September Navier-Stokes claim drew accusations over chat-log data and criticism that the model showed no work
More than 700 academics opposed OpenAI’s mathathon sponsorship; Terence Tao says mathematicians now verify AI work

Business InsiderRead more · 1 min →

Today'sHealthcare

Kevin Weil Launches Warp Sciences

Kevin Weil launched Warp Sciences to use robots for drug discovery lab work
The startup raised $215 million at an $865 million valuation, led by Lightspeed
Warp Sciences aims to automate compound testing and bring outsourced lab work back to the United States

TechCrunchRead more · 1 min →

Today's

Anthropic Reports on Hiring and AI Expectations

Anthropic published two reports comparing changes in its workforce with expectations for AI’s future.
Junior employees fell from 27% to 9% of technical hires; AI agents’ operating time reached about six times human working hours.
By 2030, employees expect AI to perform 75% of their work hours, compared with 28% among U.S. workers.

Ledge AIRead more · 3 min →

Today's

ASPR AI Launches Private Sales Agent Studio

ASPR AI launched Private Sales Agent Studio to run sales agents within customers' own cloud environments.
The company says its 11 agents cost one-tenth as much in tokens as generic AI and do not train on customer data.
New offline and online meeting notetakers support 50+ languages and send meeting outcomes or transcripts to business systems.

SalesTech StarRead more · 1 min →

Today'sFinance

Nvidia Funds European Startups That Buy Its Products

Jensen Huang visited Paris, London, Stockholm and Berlin as Nvidia invested in European startups in recent weeks
Nvidia has backed 13 European companies since the start of the year, with €83.3 million invested
Nvidia related deals with European startups totaled €5.3 billion, as some recipients buy its products and services

SiftedRead more · 1 min →

Today'sInfra

Korea-China Climate Tech: Joint Verification First

At a Seoul forum, speakers proposed starting Korea-China climate tech cooperation with joint verification rather than joint development.
They recommended combining AI and life cycle assessment, with shared standards for power mixes and emissions factors and safeguards for sensitive data.
Proposals included applying verified technologies to projects in third countries and using carbon data in green finance.

ETNewsRead more · 2 min →

Today'sRobotics

Integrate Registers Four Robot AI Patents

Integrate registers four Korean patents covering environmental perception and robot action decisions.
Company expands a predictive physical AI platform combining world models with TSFM.
Integrate plans to unveil a humanoid development kit in January 2027 with tools for development and validation.

AI Times KoreaRead more · 1 min →

Today'sSecurity

SentinelOne Reports 23-Fold Rise in AI Threat Hunting

Steve Stone said AI-assisted threat hunting improved malicious activity detection 23-fold at OneCon 2026.
SentinelOne reported a 2,200% increase over one year and an average of 28 intrusion alerts to customers daily.
The company identified attacks targeting eight customers during the first 30 days of a campaign investigation.

ETNewsRead more · 1 min →

Today'sHealthcare

Microsoft Medical AI Faces Hospital Evaluation

Microsoft says MAI-DxO scored about 83% on difficult cases, compared with about 36% for doctors.
GPT-5.5 accuracy rose from 72% to about 87% with MAI-DxO, while emergency-case results showed no clear improvement.
Microsoft plans to assess MAI-DxO’s clinical reasoning with a Boston hospital and consenting patients.

AI Times KoreaRead more · 2 min →

Today'sInfra

Korean AI Firms Step Up Japan Push

Korean AI companies showcased legal, medical, document digitization and security services at Japanese trade shows and events.
More than 80% of Japanese companies reported shortages of digital transformation staff; about 75% of one legal AI service’s users came from Japan.
Siseon AI launched a telemedicine platform in Japan, while Sparrow introduced a code security product for AI coding agents.

AI Times KoreaRead more · 2 min →

Today'sCoding

Coding AI Often Guesses at Vague DevOps Instructions

Researchers including Hong Kong University of Science and Technology tested coding AI responses to unclear DevOps instructions.
UnderspecBench tested three dimensions; DeepSeek-v4 asked questions 1.7% of the time, compared with 38.3–44.5% for Haiku 4.5.
Errors occurred in 55.8–67.8% of actions; excessive changes reached 77.2% for infrastructure settings and monitoring.

ITmedia AI+Read more · 1 min →

Today'sInfra

BIPROGY Unveils AI Platform for BankVision

BIPROGY announced an AI platform to support development and maintenance of its BankVision core banking system.
The BankVision DevAI Starter Kit was tested with 11 financial institutions in a proof of concept.
Additional testing found development productivity improved by more than 30%; SaaS service begins in November.

IT LeadersRead more · 1 min →

Trending Keywords

Last 7 Days