Category intelligence

AI News Briefing — January 29, 2026

32 current items analyzed and ranked.

Executive synthesis

AI News Summary

Model Releases & Capabilities: Moonshot AI's Kimi K2.5 claims SOTA among open models, beating Claude Sonnet 4.5 at half the cost with native multimodal understanding and 100-parallel agent swarm management. MBZUAI's K2 Think V2 advances open reasoning with a fully transparent 70B parameter pipeline. Tencent open-sourced HPC-Ops delivering 17-30% inference improvements.

Agentic AI Deployment: Google launched Auto Browse in Chrome, bringing autonomous browsing agents to billions of users. Deloitte warned that only 21% of organizations have AI agent governance despite 74% planning adoption within two years. The open-source Moltbot assistant gained 69K GitHub stars in one month despite security concerns.

Infrastructure & Geopolitics: China approved 400,000+ Nvidia H200 chips for ByteDance, Alibaba, and Tencent after weeks of blocking imports. Tesla discontinued Model S/X to pivot toward Optimus robotics. Google DeepMind unveiled AlphaGenome for analyzing 1M DNA letters to identify disease drivers. The CISA acting director accidentally leaked sensitive documents to ChatGPT, highlighting ongoing government AI security challenges.

Key Themes

Agentic AI · 8Model Releases · 4AI Infrastructure & Geopolitics · 5AI Governance & Safety · 6Enterprise AI Adoption · 5AI for Science · 3

Primary evidence

Top Ranked Signals

News Latent.Space Jan 28

[AINews] Moonshot Kimi K2.5 - Beats Sonnet 4.5 at half the cost, SOTA Open Model, first Native Image+Video, 100 parallel Agent Swarm manager

By Unknown

88 score

AI Analysis

Building on yesterday's Social buzz, Moonshot AI released Kimi K2.5, a 32B active/1T parameter MoE model claiming to beat Claude Sonnet 4.5 at half the cost while achieving SOTA on open model benchmarks. The model features native image and video understanding plus a novel 100-parallel agent swarm management capability, trained on 15T multimodal tokens.

AI News for 1/26/2026-1/27/2026. We checked 12 subreddits, 544 Twitters and 24 Discords (206 channels, and 7476 messages) for you. Estimated reading time saved (at 200wpm): 602 minutes. AINews’ website lets you search all past issues. As a reminder, AINews is now a section of Latent Space. You can opt in/out of email frequencies!Kimi has been on an absolute tear in the past year, and we last heard from them in November with Kimi K2 Thinking. Like K2, today’s K2.5 is still a 32B activ

Model ReleaseMultimodal AIAgentic AIOpen Source

News Ars Technica - All content Jan 28

Google begins rolling out Chrome's "Auto Browse" AI agent today

By Ryan Whitwam

84 score

AI Analysis

Google began rolling out 'Auto Browse,' an autonomous browsing agent integrated into Chrome that can handle tasks independently within the browser. The feature competes with OpenAI's Atlas and represents Google's most significant agentic AI deployment to date.

Google began stuffing Gemini into its dominant Chrome browser several months ago, and today the AI is expanding its capabilities considerably. Google says the chatbot will be easier to access and connect to more Google services, but the biggest change is the addition of Google's autonomous browsing agent, which it has dubbed Auto Browse. Similar to tools like OpenAI Atlas, Auto Browse can handle tedious tasks in Chrome so you don't have to. The newly unveiled Gemini features in Chrome are access

Agentic AIProduct LaunchGoogleBrowser AI

News AI (artificial intelligence) | The Guardian Jan 28

Google DeepMind launches AI tool to help identify genetic drivers of disease

By Ian Sample Science editor

82 score

AI Analysis

Google DeepMind unveiled AlphaGenome, an AI tool that can analyze up to 1 million letters of DNA code simultaneously to predict how mutations affect gene regulation. The tool could accelerate identification of genetic disease drivers and enable new treatments.

AlphaGenome can analyse up to 1m letters of DNA code at once and could pave way for new treatmentsResearchers at Google DeepMind have unveiled their latest artificial intelligence tool and claimed it will help scientists identify the genetic drivers of disease and ultimately pave the way for new treatments.AlphaGenome predicts how mutations interfere with the way genes are controlled, changing when they are switched on, in which cells of the body, and whether their biological volume controls are

AI for ScienceDeepMindHealthcare AIResearch Tool

News Ars Technica - All content Jan 28

Report: China approves import of high-end Nvidia AI chips after weeks of uncertainty

By Benj Edwards

80 score

AI Analysis

China approved imports of 400,000+ Nvidia H200 chips for ByteDance, Alibaba, and Tencent after weeks of blocking shipments despite US export clearance. The decision marks a significant shift in Beijing's stance on AI chip imports.

On Wednesday, China approved imports of Nvidia's H200 artificial intelligence chips for three of its largest technology companies, Reuters reported. ByteDance, Alibaba, and Tencent received approval to purchase more than 400,000 H200 chips in total, marking a shift in Beijing's stance after weeks of holding up shipments despite US export clearance. The move follows Beijing's temporary halt to H200 shipments earlier this month after Washington cleared exports on January 13. Chinese customs author

AI InfrastructureGeopoliticsUS-China RelationsSemiconductors

News MarkTechPost Jan 28

MBZUAI Releases K2 Think V2: A Fully Sovereign 70B Reasoning Model For Math, Code, And Science

By Maxime Mommessin

76 score

AI Analysis

MBZUAI released K2 Think V2, a fully sovereign 70B parameter open reasoning model with transparent training pipeline for math, code, and science tasks. The model uses reinforcement learning on the K2 V2 base with fully open weights and data.

Can a fully sovereign open reasoning model match state of the art systems when every part of its training pipeline is transparent. Researchers from Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) release K2 Think V2, a fully sovereign reasoning model designed to test how far open and fully documented pipelines can push long horizon reasoning on math, code, and science when the entire stack is open and reproducible. K2 Think V2 takes the 70 billion parameter K2 V2 Instruct base m

Open SourceReasoning ModelsModel ReleaseTransparency

News AI (artificial intelligence) | The Guardian Jan 28

Tesla discontinues Model X and S vehicles as Elon Musk pivots to robotics

By Nick Robins-Early

74 score

AI Analysis

Tesla announced discontinuation of Model X and Model S vehicles as Elon Musk pivots the company toward robotics, specifically the Optimus humanoid robot. The move came during earnings that beat forecasts despite declining yearly revenue.

High hopes for Optimus robot help company beat forecasts despite yearly revenue decline and flailing car businessIn the clearest sign yet that Tesla is pivoting away from its electric car business, CEO Elon Musk announced on Wednesday’s investor call that the company would discontinue production of its Model X SUV and Model S full-size sedan.“It’s time to basically bring the Model S and X programs to an end,” Musk said. “We expect to wind down S and X production next quarter.” Continue reading..

RoboticsTeslaCorporate StrategyHumanoid Robots

News AI News Jan 28

Deloitte sounds alarm as AI agent deployment outruns safety frameworks

By David Thomas

70 score

AI Analysis

Deloitte report warns that only 21% of organizations have implemented governance for AI agents despite 23% currently using them and 74% expecting to within two years. The gap between deployment speed and safety frameworks poses serious security and accountability risks.

A new report from Deloitte has warned that businesses are deploying AI agents faster than their safety protocols and safeguards can keep up. Therefore, serious concerns around security, data privacy, and accountability are spreading. According to the survey, agentic systems are moving from pilot to production so quickly that traditional risk controls, which were designed for more human-centred operations, are struggling to meet security demands. Just 21% of organisations have implemented stringe

AI GovernanceAgentic AIEnterprise AIAI Safety

News Ars Technica - All content Jan 28

US cyber defense chief accidentally uploaded secret government info to ChatGPT

By Ashley Belanger

68 score

AI Analysis

The acting director of CISA accidentally uploaded sensitive government contracting documents to public ChatGPT, triggering internal cybersecurity warnings. The incident occurred despite DHS restrictions blocking most staff from using OpenAI's chatbot.

Alarming critics, the acting director of the Cybersecurity and Infrastructure Security Agency (CISA), Madhu Gottumukkala, accidentally uploaded sensitive information to a public version of ChatGPT last summer, Politico reported. According to "four Department of Homeland Security officials with knowledge of the incident," Gottumukkala's uploads of sensitive CISA contracting documents triggered multiple internal cybersecurity warnings designed to "stop the theft or unintentional disclosure of gove

AI SecurityGovernment AIPolicyData Privacy

News Ars Technica - All content Jan 28

Users flock to open source Moltbot for always-on AI, despite major risks

By Benj Edwards

65 score

AI Analysis

First spotted on Social, now making mainstream headlines, Moltbot, an open source AI assistant formerly called Clawdbot, gained 69,000 GitHub stars in one month, making it one of 2026's fastest-growing AI projects. The tool enables proactive AI communication across messaging platforms but carries significant security risks.

An open source AI assistant called Moltbot (formerly "Clawdbot") recently crossed 69,000 stars on GitHub after a month, making it one of the fastest-growing AI projects of 2026. Created by Austrian developer Peter Steinberger, the tool lets users run a personal AI assistant and control it through messaging apps they already use. While some say it feels like the AI assistant of the future, running the tool as currently designed comes with serious security risks. Among the dozens of unofficial AI

Open SourceAI AssistantsConsumer AIAI Security

News AI (artificial intelligence) | The Guardian Jan 28

Microsoft shrugs off AI bubble fears again with strong financial results

By Edward Helmore

62 score

AI Analysis

Microsoft reported Q2 revenues of $81.27 billion with strong results despite slowing cloud growth, demonstrating continued AI boom momentum. The results counter AI bubble concerns and support ongoing AI investment.

Company reports second-quarter revenues of $81.27bn but posts slowing growth in key cloud computing businessInvestor interest in Microsoft shares may have weakened in recent months, but the company posted strong financial results on Wednesday which yet again demonstrated that the AI boom is roaring on.Microsoft reported earnings for the second quarter of fiscal year that are likely to keep the party going for Wall Street, despite slowing growth in its key cloud computing business. Continue readi

Enterprise AIBusinessCloud ComputingMicrosoft

News Feed: Artificial Intelligence Latest Jan 28

Google’s New Chrome ‘Auto Browse’ Agent Attempts to Roam the Web Without You

By Reece Rogers

62 score

AI Analysis

Google's Auto Browse feature puts generative AI in control of web navigation, allowing users to delegate browsing tasks. The feature represents Google's response to competing agentic products.

Google’s latest addition to its Chrome browser puts generative AI behind the wheel and you in the passenger seat.

Agentic AIGoogleBrowser AIProduct Launch

News MarkTechPost Jan 28

Tencent Hunyuan Releases HPC-Ops: A High Performance LLM Inference Operator Library

By Michal Sutter

62 score

AI Analysis

Tencent Hunyuan open-sourced HPC-Ops, a production-grade operator library for LLM inference delivering ~30% QPS improvement for Tencent-HY models and ~17% for DeepSeek models. The library provides optimized CUDA kernels for attention, GEMM, and MoE operations.

Tencent Hunyuan has open sourced HPC-Ops, a production grade operator library for large language model inference architecture devices. HPC-Ops focuses on low level CUDA kernels for core operators such as Attention, Grouped GEMM, and Fused MoE, and exposes them through a compact-C and Python API for integration into existing inference stacks. HPC-Ops runs in large scale internal services. In those deployments it delivers about 30 percent queries per minute improvement for Tencent-HY models and

Open SourceAI InfrastructureInference OptimizationTencent