Latest AI News

Get the fastest, most accurate AI news every morning

Total articles: 12,861|Added todayToday: 91

Category

Search

Apple Launches Japanese Siri AI Beta

Apple launched the Japanese-language Siri AI beta on October 26, 2026.
The beta works across five operating systems and supports screen understanding and actions across multiple apps.
Some features have daily usage limits; Apple plans paid access with expanded limits.

Ledge AIRead more · 1 min →

Today'sEducation

Students Urge Schools to Teach AI Literacy

Project Tomorrow’s 2025-26 report draws on more than 95,000 people across 1,023 schools
Forty-six percent of high school students learned about AI independently; 30 percent learned in class
Students urged schools to teach AI literacy and connect lessons with real-world skills

GovTech AIRead more · 2 min →

Today'sSecurity

SCSK Security Announces AI SOC Service

SCSK Security announced its AI SOC service on October 28, 2026.
The service uses existing security platforms to monitor and analyze every alert around the clock.
In an internal test, it averaged two minutes per initial analysis and automatically closed about 95% of alerts.

IT LeadersRead more · 1 min →

Today'sSecurity

IBM Japan Outlines Vulnerability Management for the AI Era

IBM Japan’s Tetsuya Kono outlined vulnerability management from development through production at Impress AI Days 2026.
Time to exploit is forecast to fall to 1 minute in 2028; Kono said some vulnerabilities are already exploited in an average of 29 minutes.
IBM Concert Protect combines scan findings with system context to calculate risk and prioritize fixes.

Impress DIGITAL XRead more · 2 min →

Hitachi Solutions Launches Opnova AI Agent

Hitachi Solutions began offering Opnova, a PC automation tool developed by US-based Opnova, on October 29.
The AI agent automates workflows using screen recognition, recorded user actions and natural-language instructions.
A European retailer with more than 5,700 stores saves over 3,000 work hours annually by automating invoice checks.

IT LeadersRead more · 1 min →

Meta and Harvard Researchers Study AI Teamwork

Meta FAIR, Harvard researchers and others proposed ScaleMAS to coordinate multiple AI agents.
On software repair, four ScaleMAS agents matched eight independent agents using nearly half the tokens.
Document contradiction detection improved through 128 agents, while theorem proving plateaued at about 16.

ITmedia AI+Read more · 1 min →

KDDI Iret Launches 113 Free AI Case Studies

KDDI Iret published a free site featuring 113 examples of AI use inside the company.
The collection includes 63 engineering cases and 50 cases from sales, HR, and other non-engineering roles.
Examples include checking approval-request data with Gemini Notebook and using Gemini to gather cases and prepare materials.

ITmedia AI+Read more · 1 min →

Trending Keywords

Liner Router API Tested on Kubernetes Incidents

Liner’s Router API scored near GPT-6.1 Sol on 10 unpracticed Kubernetes incidents, with a 0.02 operating-score gap.
Liner cost 28% of GPT-6.1 Sol but 3.3 times as much as lightweight GPT-6 Luna.
Blind review matched scored accuracy and risk-action judgments in 89 of 90 runs.

Yozm ITRead more · 3 min →

Silicon Photonics and CPO Target AI Data Bottlenecks

SK hynix Tech Note 3 explains AI factory data bottlenecks and silicon photonics and CPO.
In Meta’s 2025 evaluation, CPO optical links used about 65% less power than the pluggable comparison.
HBM supports the GPU memory path, while CPO improves network efficiency between GPU systems.

SK HynixRead more · 3 min →

Sakana AI Opens Early Access to Enterprise API Platform

Sakana AI announced its enterprise API platform on October 27, 2026, and began accepting early access applications ahead of general availability.
The platform will initially offer Sakana Fugu and Sakana Fugu Ultra; Sakana Namazu and Sakana Fugu Max are planned for later release.
Enterprise features include organization management, access controls, audit logs, budget limits, and centralized usage tracking.

Sakana AIRead more · 2 min →

Paper Finds Memory Evolution Outperforms Test-Time Training

Research paper introduces Test-Time Memory Evolution, which updates external memory without changing model weights.
In a shared evaluation framework, TTME outperformed test-time training on all five mathematical tasks.
TTME reduced FLOPs by 82.8-84.6% and runtime by 47.7-58.4% in evaluated settings.

HuggingFaceRead more · 1 min →

How Work Is Redesigned for AI Collaboration

SK Telecom CEO Jae-heon Jeong asked what people should build with AI at a September 1 lecture.
A study of 244 BCG consultants found 60% were “cyborgs,” 14% “centaurs” and 27% “self-automators.”
About 80% followed AI’s wrong answers; researchers stressed defining problems, monitoring work and checking results.

SK Telecom NewsroomRead more · 3 min →

OpenRouter Rebuilds Provider Routing

OpenRouter announced a default provider-routing algorithm that weighs price, speed, reliability and response quality.
A one-week test across more than 24 million requests found 16.9% lower median TTFT and 16.3% higher throughput.
The test recorded 67% fewer requests hitting HTTP 429 errors, while cost per 1,000 requests rose 6.0%.

OpenRouterRead more · 2 min →

Yesterday'sFinance

Microsoft Revenue Rises 19% on Cloud Growth

Microsoft’s fiscal first-quarter 2027 revenue rose 19% to $92.279 billion, topping estimates.
Azure revenue surged 48% to $33.17 billion, while commercial RPO grew 73%.
Capital spending fell short of the outlook, and free cash flow reached $18.6 billion, more than double estimates.

Aju Business DailyRead more · 2 min →

Anthropic IPO Filing Raises Mission Questions

Anthropic’s IPO prospectus states its public-benefit purpose while warning that advanced AI could pose catastrophic or existential risks.
First-half 2026 revenue was $16.3 billion; operating expenses neared $9 billion and non-cancellable commitments totaled roughly $500 billion.
Seven founders retain 50.1% voting power through Class F shares, while a trust can elect a majority of the board.

latimes.comRead more · 2 min →

Google’s Gemini 4 Carbon Tops AI Benchmark

Gemini 4 Carbon ranked first among 52 models on AAII with a score of 59.
The model scored 59 on a coding-agent index and posted top-tier results in cybersecurity evaluations.
Google is giving Carbon to selected security experts first; a general release date has not been announced.

AI Times KoreaRead more · 3 min →

Speakers Outline Human-AI Teamwork

Speakers at Connected Enterprise 2026 called for smaller teams combining people and AI agents.
AI adds an average of three hours to team workweeks, Ron Friedman said, citing extra content and coordination.
Karthik Krishnamurthy proposed 50% base pay, 50% variable pay and project profit sharing for engineers.

Constellation ResearchRead more · 1 min →

Yesterday'sInfra

Banks Use Smaller AI Models to Manage Costs

Bank of America and Capital One executives described model routing as a way to manage AI costs at the Evident AI Symposium.
Bank of America plans to double its AI budget next year and uses Orchestra to route queries to cost-effective models.
Capital One says smaller models cut latency by five times and raised dealership customer engagement by 50%.

CIO DiveRead more · 1 min →

Yesterday'sHealthcare

Counsel Health Launches AI Weight Program

Counsel Health and Eli Lilly launched an AI-supported weight management program through LillyDirect and Counsel Health.
Members in 44 states can access physicians on demand; annual unlimited membership costs $300 or $29 per visit.
GLP-1 use for weight loss reached 11% of U.S. adults in 2026, while therapy retention falls to 15% by year two.

Fierce HealthcareRead more · 1 min →

Cohere Details AWS Financial Research Assistant

Cohere outlines an AWS financial research assistant that answers filing questions with page-level citations
Embed Pro indexes pages while Embed Fast processes questions; 1,024-dimension int8 vectors cut raw storage 8x
The demo combines Rerank and selective Parse, with adjustable stages, relevance scores, and searchable page images

CohereRead more · 2 min →

Trending Keywords

Last 7 Days