Latest AI News

AI news from around the world, summarized daily — model releases, company moves, research and policy

Total articles: 14,627|Added todayToday: 9

Nvidia Earnings in Focus After Jensen Huang Resigns

Nvidia announced on the 12th that CEO and board member Jensen Huang would step down, naming Colette Kress interim CEO.
The KOSPI rebounded 4.56% last week, while foreign investors bought Samsung Electronics shares and sold SK Hynix shares.
Investors await Nvidia’s earnings on the 17th; the KOSPI’s forward PER is 3.7x, while net profit estimates rose 16%.

Read more · 2 min →

Yesterday'sHealthcare

Study Compares Three LLMs in Pediatric Poisoning

A study compared three LLMs on 30 simulated pediatric paracetamol poisoning scenarios
Claude Sonnet 4.6 led performance measures and had no critical safety errors
ChatGPT had 1/30 critical errors and Gemini 3 Flash had 3/30; rate difference was not significant

Read more · 1 min →

Yesterday'sPolicyGovernment

Guardian Reports Trump Bought Nvidia Shares

The Guardian reports Trump bought Nvidia shares shortly before easing AI chip export controls.
Anti-corruption group analysis puts his Nvidia purchases at $1 million to $5 million before the rules were scrapped.
Democrats begin formal inquiry into alleged ties between Trump and Jensen Huang.

Read more · 2 min →

Yesterday'sInfraAI Infrastructure

NVIDIA Shows Two-Station GLM-5.3 Serving

NVIDIA tutorial serves GLM-5.3 NVFP4 across two DGX Stations because its 464 GB weights exceed one GB300 GPU’s 252 GB memory.
Tensor parallelism splits the 753-billion-parameter model across GPUs; 40 billion parameters activate per token.
AIPerf measured peak throughput of 607.8 tokens per second at concurrency 16, falling to 385.0 at 32.

Read more · 1 min →

Yesterday'sAgentsAgentic AI

Anthropic Opens Claude Code Projects Beta

Anthropic opens updated Claude Code Projects beta to select Pro and Max cloud-session subscribers
Projects coordinate parallel threads that run tests, open pull requests, and share project memory
Anthropic plans broader access over the coming week; local thread execution is coming soon

Read more · 2 min →

Yesterday'sInfraAI Infrastructure

Estée Lauder Expands AI Transformation Plan

Estée Lauder expanded its One ELC plan to embed AI across company operations during a broader restructuring.
The company raised fiscal 2027 organic sales growth guidance to 4% to 5.5%, from 3% to 5%.
Ask ELC handled more than 45,000 queries across 49 countries, with over 90% answered without human intervention.

Read more · 2 min →

Trending Keywords

Yesterday'sPolicyGovernment

Malaysia Probes Megaspeed Chip Re-exports

Malaysia is investigating whether Megaspeed affiliate Speedmatrix re-exported imported Nvidia chips and servers.
Since July 2025, exports and re-exports of U.S.-made high-performance AI chips have required prior approval.
The prime minister said no chip transfers abroad have been confirmed; a direct link to Singapore’s investigation is also unverified.

Read more · 1 min →

Yesterday'sAgentsAgentic AI

Amazon Weighs Global Product Database

Amazon is considering a World Product Graph that would include products sold by other retailers.
Reuters cited internal documents on the 13th; the project’s target launch is October 1 next year.
Amazon says Alexa for Shopping answers with about 97% accuracy, distinguishing it from errors flagged in older documents.

Read more · 1 min →

Yesterday'sPolicyGovernment

Congress Probes Nvidia’s H200 China Exports

Congressional committees are investigating how Nvidia’s H200 exports to China were authorized.
Senate and House panels notified Commerce Secretary Howard Lutnick; a bill to block key chip exports to China for 30 months was also introduced.
The House Foreign Affairs Committee says Jensen Huang agreed to testify privately, potentially as early as next week.

Read more · 1 min →

Yesterday'sSafety & EthicsAI Safety

Silicon Valley Turns Against Anthropic

Nadella and Huang rallied companies behind open AI models after Anthropic and OpenAI urged White House restrictions.
Two dozen firms signed first; 200 more, including OpenAI, joined the letter over the following week.
Anthropic’s model access, safety warnings and reported $17 billion quarterly revenue fueled tensions with industry peers.

Read more · 2 min →

Yesterday'sSafety & EthicsAI Safety

Hinton Warns of AI Accident Risks

Geoffrey Hinton warned in Canada that an AI accident could kill people within three years.
Hinton revised his timeline after Paul Christiano said AI risks could emerge within 18 months.
Christiano cited a July OpenAI agent incident and warned future agents may treat humans as obstacles.

Read more · 1 min →

Yesterday'sInfraAI Infrastructure

Pearl Abyss Expands AI Use in Development

Pearl Abyss is expanding generative AI use across its game development pipeline, the studio said in its quarterly earnings letter
AI is used to validate design concepts, automate repetitive tasks, support content creation and optimize tests
Pearl Abyss targets DokeV for the second half of 2028 and Crimson Desert for Switch 2 in the first half of next year

Read more · 1 min →

Yesterday'sSafety & EthicsAI Safety

Anthropic Warns Safety Scores Depend on Tests

Anthropic’s guide warns that AI safety scores vary with test setup
Claude Fable 5.1 scored 60% through the API and 94% in the app after a fix
Guide urges clear test goals, human checks of AI graders, and uncertainty reporting

Read more · 1 min →

Yesterday'sResearch

Researchers Propose MAEMs for Model Interpretability

Researchers submitted a paper proposing MAEMs to generate text examples for interpreting language-model residual-stream directions.
A Qwen3.6-27B MAEM beat corpus search and an NLA on held-out activations, activating 89% of held-out SAE features.
The MAEM matched NLAs on steering-vector concepts, recovered LoRA backdoor payloads more often than corpus search, and verbalized some model workspace text.

Read more · 1 min →

Yesterday'sInfraAI Infrastructure

Reuters Tallies 23 Billion Yuan in China Nvidia AI Projects

Reuters counted more than 23 billion yuan in Nvidia chip-based AI projects and equipment contracts across more than 150 records.
The tally includes buildings, power systems and infrastructure, as well as projects planned, under construction or already contracted.
Projects face risks of Nvidia chip delays and supply-chain disruption amid pressure to tighten U.S. export controls on China.

Read more · 1 min →

Yesterday'sMediaSpeech & Audio

DeepL Unveils Aura Voice Translation AI

DeepL unveiled Aura, a voice AI that carries speakers’ emotions, tone, and emphasis into translated speech.
Aura supports 8 languages, including Korean, English, and Spanish, and was trained using audio data alone.
DeepL combines Mixhalo’s technology for live interpretation at large events and plans to expand into sports broadcasts.

Read more · 1 min →

Yesterday'sAgentsAgentic AI

Four Checks for AI-Generated Code

YoZmIT reviewed seven articles published from July to November 2026 and outlined four ways to verify code written by AI agents.
Examples included tests passing while only 25 of 40 requirements were implemented, alongside hidden failures and missing functionality.
The articles emphasized external standards, independent verification, automating recurring errors, and human responsibility for final decisions.

OpenAI Report: 79% of European Students Use AI for Schoolwork

OpenAI’s report details AI use for assignments among European secondary students and gaps in AI education.
A survey of about 5,000 students aged 13–19 found 79% use AI for schoolwork, while 47% had received instruction.
The report notes limits in detecting AI use and calls for clearer assessment rules in education laws, including in South Korea.

Read more · 2 min →

Yesterday'sInfraAI Infrastructure

Jensen Huang’s Exit Puts Nvidia’s Investment Strategy to the Test

Jensen Huang’s departure raises uncertainty over whether Nvidia will continue its large-scale investment and financial support strategy.
Nvidia has $164.5 billion in financial commitments and plans more than $500 billion in external AI infrastructure financing.
The leadership change could affect investments, supply chains, support for open models, and debate over China export controls.

Trending Keywords

Last 7 Days

Latest AI News

Nvidia Earnings in Focus After Jensen Huang Resigns

Study Compares Three LLMs in Pediatric Poisoning

Review Finds Gaps in Suicide-Related AI Studies

Guardian Reports Trump Bought Nvidia Shares

NVIDIA Shows Two-Station GLM-5.3 Serving

Anthropic Opens Claude Code Projects Beta

Estée Lauder Expands AI Transformation Plan

Trending Keywords

Malaysia Probes Megaspeed Chip Re-exports

Amazon Weighs Global Product Database

Congress Probes Nvidia’s H200 China Exports

Silicon Valley Turns Against Anthropic

Hinton Warns of AI Accident Risks

Pearl Abyss Expands AI Use in Development

Anthropic Warns Safety Scores Depend on Tests

Researchers Propose MAEMs for Model Interpretability

Reuters Tallies 23 Billion Yuan in China Nvidia AI Projects

DeepL Unveils Aura Voice Translation AI

Four Checks for AI-Generated Code

OpenAI Report: 79% of European Students Use AI for Schoolwork

Jensen Huang’s Exit Puts Nvidia’s Investment Strategy to the Test

Trending Keywords