All Updates

Page 694 of 754

February 19, 2026

🏠
IT之家59d ago

Microsoft Splits Win11 Canary into Dual Tracks

Microsoft split the Windows 11 Canary channel into 28000 series for 26H2 feature previews and new 29500 series for platform development. The 29500 track targets kernel, drivers, and hardware compatibility with high instability risks. Upgrading is one-way, requiring clean install to revert.

#insider-program#kernel-dev#hardware-compat
🦊
GitLab Blog59d ago

GitLab Security Dashboard Adds Remediation Trends

GitLab 18.9 updates the Security Dashboard with trend tracking, vulnerability age distribution, and risk scoring. It helps prioritize remediation using factors like EPSS and KEV scores. Teams can monitor velocity, identify training needs, and reduce manual reporting.

#risk-scoring#devsecops
🦊
GitLab Blog59d ago

GitLab Exposes North Korean Dev Scams

GitLab's Threat Intelligence Team published a report detailing North Korean actors' use of the platform for Contagious Interview malware campaigns and IT worker fraud since 2022. They banned offending accounts in 2025 and shared insights into malware repos, synthetic identities, and financial records. No customer action required; GitLab remains secure.

#threat-intelligence#nation-state#malware
🦊
GitLab Blog59d ago

GitLab Duo Self-Hosted and BYOM Launch

GitLab 18.9 enables self-hosted Duo Agent Platform for online cloud licenses with usage-based GitLab Credits billing. It supports data residency and control for regulated industries. Bring Your Own Model (BYOM) integrates custom models via GitLab AI Gateway for full enterprise governance.

#agentic-ai#data-residency#self-hosted
🤝
Together AI Blog59d ago

CDLM: 14x Faster Diffusion LM Inference

Together AI's Consistency Diffusion Language Models (CDLM) overcome standard diffusion LM limitations by enabling exact block-wise KV caching and trajectory-consistent step reduction. This post-training recipe delivers up to 14.5x latency improvements without quality loss. It makes diffusion LMs practical for real-world use.

#diffusion-models#kv-caching#step-reduction

February 18, 2026

🏠
IT之家59d ago

Sony Monetizes PS5 Users to Dodge Memory Hikes

Sony plans to offset AI-driven memory price surges by monetizing 92M existing PS5 users via software/network services. Hardware prices stay stable, but PS Plus hikes likely. PS6 delayed to 2028-2029 awaiting market normalization.

#memory-shortage#ai-chip-crisis#subscription-model
☁️
AWS Machine Learning Blog59d ago

AWS Bedrock AgentCore Enables Unified Intelligence

AWS Machine Learning Blog post demonstrates building unified intelligence systems with Amazon Bedrock AgentCore. It showcases a real-world implementation called Customer Agent and Knowledge Engine (CAKE). This helps integrate agents and knowledge bases effectively.

#agent#knowledge-engine#unified-intelligence
💾
PCMag60d ago

Microsoft AI Inserts Typos in Old Diagram

A programmer's 15-year-old diagram ended up on Microsoft Learn with embarrassing typos apparently added by AI. Vincent Driessen highlighted the curious errors generated during processing. This reveals potential pitfalls in AI content handling.

#ai-hallucination#technical-docs#content-errors
📊
Bloomberg Technology60d ago

Palantir Partial Win vs Ex-Workers' AI Startup

A judge sided with Palantir, ruling three ex-employees likely violated confidentiality and non-solicitation agreements by founding AI startup Percepta. However, the court declined to halt their operations at the new company.

#non-compete#lawsuit#talent-mobility
🏠
IT之家60d ago

Meta AI Smartwatch Launch Planned for 2025

Meta plans to launch an AI-enabled smartwatch with health tracking later this year, reviving a canceled project codenamed Malibu 2. It precedes delayed Phoenix MR glasses pushed to 2027. Faces competition from Apple, Google, and others in wearables.

#wearables#health-tracking#gesture-control
🗾
ITmedia AI+ (日本)60d ago

KnowBe4's Deepfake Boss Videos for Phishing Training

KnowBe4 launched a Japanese version of training using deepfake videos and audio of its executives to simulate sophisticated impersonation attacks. The program allows users to experience advanced AI scams hands-on. It aims to sharpen judgment skills against AI-driven fraud.

#deepfake#phishing-training#security-simulation
🗾
ITmedia AI+ (日本)60d ago

Japan IBM's AI Strategy: 3 Pillars for Legacy Modernization

Japan IBM announced its AI strategy, positioning 2026 as the inaugural year for full AI integration into system development and business processes. The company is launching an AI platform equipped with development support tools and governance features. It emphasizes strengths in hybrid environments for large enterprises.

#hybrid-cloud#governance#legacy-modernization
📰
The Verge60d ago

Meta Plans AI Smartwatch Launch This Year

Meta is reportedly set to launch the Malibu 2 smartwatch with health tracking and AI features later this year, ahead of updated Meta Ray-Ban smart glasses. The company has delayed its Phoenix mixed reality glasses until 2027 to streamline its AR/MR roadmap. This follows the cancellation of prior smartwatch plans in 2022 due to technical and cost issues.

#wearables#health-tracking#ar-mr
🇬🇧
BBC Technology60d ago

48-Hour Mandate for Removing Abusive Images

The government proposes a new law requiring tech firms to remove abusive images within 48 hours. Intimate image abuse will be treated as a more severe offense. This aims to strengthen protections against image-based sexual abuse.

#content-moderation#uk-regulation#image-abuse
🗾
ITmedia AI+ (日本)60d ago

Microsoft Flags AI Memory Poisoning Attacks

Microsoft warns of 'AI Recommendation Poisoning,' a technique abusing generative AI memory functions via malicious URLs to bias recommendations toward specific companies. Over 50 cases have been confirmed, posing a new threat to AI neutrality.

#prompt-injection#memory-exploit
🇨🇳
cnBeta (Full RSS)60d ago

WD Invests $74M in Thailand HAMR R&D

Western Digital approved for 23 billion THB (~$73.8M) investment in Thailand HAMR HDD R&D center. The project expands existing facilities to advance heat-assisted magnetic recording tech. Aims to boost local HDD production capabilities and value.

#hdd-storage#thailand-boost#areal-density
📊
Bloomberg Technology60d ago

TPG CEO on AI Opportunities Amid IPO Uncertainty

TPG CEO Jon Winkelried discussed opportunities in AI, worries around software investments, and uncertainty in IPOs during a Bloomberg Deals interview. TPG manages over $300 billion across diverse strategies.

#ai-disruption#ipo-uncertainty#software-investments
💼
VentureBeat60d ago

GEA Matches Human AI Agents at Zero Cost

UC Santa Barbara researchers developed Group-Evolving Agents (GEA), enabling AI agent groups to evolve collectively by sharing experiences and innovations. GEA outperformed existing self-improving frameworks on coding and software tasks, matching or exceeding human-engineered systems. It eliminates inference costs and overcomes individual-centric evolution silos.

#agent-evolution#zero-inference
📡
AI Wire60d ago

DOE's 26 AI Challenges for Genesis Mission

The Department of Energy (DOE) has outlined 26 key AI challenges for the Genesis Mission, emphasizing breakdowns in scientific AI at the data layer. Fragmented datasets are highlighted as a primary issue. AIwire details these challenges that will shape the mission's success.

#data-layer#fragmented-datasets#scientific-ai
📊
Bloomberg Technology60d ago

Pritzker Halts Data Center Tax Perks Amid Power Surge

Illinois Governor JB Pritzker seeks to temporarily halt tax incentives for data centers. The move aims to contain soaring power bills driven by high energy consumption. Data centers are expanding rapidly due to AI demands.

#tax-incentives#power-bills#policy-halt
Page 694 of 754