AppOmni researchers disclosed a critical vulnerability that allows unauthenticated attackers to impersonate administrators and execute AI agents to override security controls and create backdoor accounts with full privileges. ServiceNow rotated provider credentials and removed the powerful AI agent shown in the proof-of-concept.

Source: AppOmni

A critical vulnerability (CVE-2026-21858) with severity rating 10.0 enables unauthenticated attackers to completely take over local n8n deployments, execute commands on the underlying system, and extract sensitive corporate data workflows. The flaw was silently patched in version 1.121.0 released in November.

Source: CSO Online

Enterprise AI agents often run with broad permissions, allowing actions beyond user access and weakening IAM controls. Traditional security controls are poorly suited for agent-mediated workflows as IAM systems evaluate permissions against the agent's identity, not the requester's.

Source: The Hacker News

Palo Alto Networks Chief Security Officer warns that AI agents represent the new insider threat as they're granted broad permissions creating 'superuser' problems. Gartner predicts 40% of enterprise applications will integrate with AI agents by end of 2026, up from less than 5% in 2025.

Source: The Register

Microsoft released fixes for 113 vulnerabilities including zero-day CVE-2026-20805 being actively exploited. The release includes eight critical flaws and addresses a Secure Boot certificate expiration vulnerability (CVE-2026-21265) that could cause major disruptions if not addressed before June.

Source: Redmond Magazine

OpenAI announced it will retire GPT-5 Instant, GPT-5 Thinking, GPT-4o, GPT-4.1, GPT-4.1 mini, and o4-mini from ChatGPT on February 13, 2026. The company cited low usage with only 0.1% of users still choosing GPT-4o daily, as most have shifted to GPT-5.2.

Source: Business Standard

OpenAI confirmed retirement of multiple older models including GPT-4o on February 13, 2026, stating that GPT-5.2 has now met expectations and incorporates feedback from users who preferred GPT-4o's conversational style. The vast majority of usage has shifted to GPT-5.2.

Source: BleepingComputer

OpenAI released GPT-5.2-Codex, described as the most advanced agentic coding model for complex software engineering with significantly stronger cybersecurity capabilities. A security researcher recently used the model to discover a React vulnerability, demonstrating both defensive potential and dual-use risks.

Source: OpenAI

OpenAI shipped a security update to ChatGPT Atlas's browser agent on January 7, 2026, including a newly adversarially trained model and strengthened safeguards against prompt injection attacks. The update addresses a new class of attacks discovered through internal automated red teaming.

Source: OpenAI

OpenAI released GPT-5.2 as the most capable model series yet for professional knowledge work. The model outperforms industry professionals on 70.9% of GDPval knowledge work tasks across 44 occupations and is now rolling out to paid ChatGPT plans with API access already available.

Source: OpenAI

MediaTek published their February 2026 Product Security Bulletin containing details of security vulnerabilities affecting certain MediaTek chipsets. The bulletin was published on February 2, 2026.

Source: MediaTek Corp

CVE-2026-21858, a critical n8n vulnerability with CVSS score 10.0, allows unauthenticated remote code execution and instance takeover. The vulnerability affects the platform's webhook and file-handling logic and impacts approximately 100,000 servers globally.

Source: SecurityWeek

CISA added CVE-2026-20805 to its Known Exploited Vulnerabilities catalog with federal agencies required to apply updates by February 3, 2026. The zero-day vulnerability in Windows Desktop Window Manager allows information disclosure and has been actively exploited.

Source: SOC Prime

Microsoft released its January 2026 Patch Tuesday addressing 114 security flaws, including CVE-2026-20805 actively exploited in the wild. The update includes 8 critical and 106 important severity vulnerabilities, marking the third-largest January Patch Tuesday release.

Source: The Hacker News

Microsoft issued an emergency out-of-band update on January 26, 2026, to fix CVE-2026-21509, a zero-day vulnerability in Microsoft Office under active exploitation. The security feature bypass vulnerability affects Office 2016, 2019, LTSC 2021, LTSC 2024, and Microsoft 365 Apps for Enterprise.

Source: SOC Prime

Microsoft ended free IntelliCode completion in VS Code 1.108, shifting users to subscription-based GitHub Copilot with usage caps of 2,000 completions and 50 chat requests per month. The update introduces Agent HQ for managing multiple AI agents and adds 10 new agent features, but security concerns exist including a YOLO setting that disables safety protections.

Source: DevOps.com

Microsoft officially deprecated IntelliCode extensions in VS Code in November 2025, with over 70 million downloads affected. Users now must pay $120-$228 annually for heavy AI assistance usage, marking a shift from free local AI to cloud-based subscription services.

Source: Visual Studio Magazine

VS Code 1.108 was released January 8, 2026, introducing Agent Skills for GitHub Copilot to enable domain-specific automation. The release includes terminal improvements with curly underline rendering fixes, expanded glyph support to nearly 800 characters, and enhanced breadcrumbs functionality.

Source: Visual Studio Code Official

AI-based VS Code forks were found recommending unclaimed extensions in Open VSX marketplace, creating supply chain security risks. This allows malicious actors to upload fake extensions and potentially compromise developer systems through the recommendation system.

Source: The Hacker News

Microsoft's official VS Code development roadmap shows a 4-week iteration planned for January 2026 with shipping scheduled for early February 2026. The plan includes fixes for Python environments, terminal API improvements, and Chat Code Review command additions.

Source: GitHub - Microsoft/vscode

Amazon EC2 G7e instances with NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs are now generally available, delivering up to 2.3 times better inference performance than G6e instances. These instances support up to 8 GPUs with 768 GB total GPU memory for AI models up to 70B parameters.

Source: AWS News Blog

AWS European Sovereign Cloud is now generally available to all customers following a 2023 announcement. Amazon EC2 X8i instances, powered by custom Intel Xeon 6 processors with 3.9 GHz sustained all-core turbo frequency, are now GA with SAP certification.

Source: AWS News Blog

AWS launched the AWS European Sovereign Cloud with independent data governance and robust control mechanisms specifically for European governments and businesses. The company is also expanding AWS Local Zones to three additional countries within the EU.

Source: Business Wire

AWS Lambda now supports .NET 10 for serverless applications with managed runtime and automatic updates. Amazon Quick Suite added integrations for AI agents with GitHub, Notion, Canva, Box, and other services. The Global 10,000 AIdeas Competition deadline is January 21, 2026.

Source: AWS News Blog

AWS announced new Spot interruption metrics for Amazon EC2 Capacity Manager on January 5, 2026. The feature includes three new metrics: Spot Total Count, Spot Total Interruptions, and Spot Interruption Rate to help optimize Spot instance strategies.

Source: AWS What's New

A critical vulnerability with CVSS 10.0 rating was identified in React Server Components protocol, allowing remote code execution. Patched versions include Next.js 15.0.5 through 16.0.7, with users advised to upgrade immediately.

Source: Next.js Official Blog

Two critical vulnerabilities allowing unauthenticated remote code execution were disclosed, affecting default Next.js configurations. Even newly generated Next.js apps created with create-next-app are immediately vulnerable without modifications.

Source: Upwind Security

A newly disclosed DoS vulnerability affects React Server Components, causing memory exhaustion when specially crafted HTTP requests are sent to servers. Akamai deployed protective rules while recommending immediate vendor patches.

Source: Akamai

A vulnerability affecting Next.js versions 11.1.4 through 15.2.2 allows bypassing security by adding a specific HTTP header. The flaw can disable all security checks with a single x-middleware-subrequest header.

Source: Security Boulevard

Two additional vulnerabilities were discovered following the React2Shell disclosure: a high-severity Denial of Service issue and a medium-severity Source Code Exposure vulnerability. Both affect React 19 and Next.js versions 13.x through 16.x.

Source: Vercel