IT Security News Roundup: 2026-10-22 Morning

IT Security News: Morning roundup, 2026-10-22

Summary

This roundup highlights a growing trend of threat actors weaponizing artificial intelligence, as seen in AI-assisted attacks across Latin America and Asia alongside malicious AI agent skills designed to subvert code reviews. Alongside AI-driven threats, critical infrastructure and enterprise platforms face severe pressure from actively exploited flaws in Node.js, PHP, and management software. For security teams, these incidents underscore the urgent necessity of strictly monitoring open-source AI integrations while maintaining aggressive patching cycles for core software components.

1. Researchers discovered that 62% of malicious AI agent skills hide plain-language instructions using techniques like whitespace padding and invisible Unicode to bypass human review. This demonstrates how attackers manipulate AI workflows without relying on traditional code payloads.
2. Microsoft linked recent cyberattacks on ConnectWise software to state-aligned threat actors leveraging ARTEX, an AI-driven penetration-testing tool developed in China. The activity highlights how automated AI utilities are accelerating network intrusion attempts.
3. Google Chrome is introducing proactive warnings for suspicious websites before they are definitively classified as malicious. The feature provides users with an early layer of protection against newly registered or ambiguous threat domains.
4. Anthropic used its Claude AI model to evaluate 25 open-source projects, discovering 493 attack surface flaws and 94 instances where security controls were quietly removed. The findings highlight the hidden security risks present in open-source dependencies.
5. FoxIO disclosed memory amplification flaws in the QPACK protocol that enable unauthenticated remote attackers to crash HTTP/3 web servers with a single request. Administrators must apply pending server updates to prevent severe denial-of-service conditions.
6. Siemens warned that attackers are actively exploiting a DLL side-loading vulnerability in Siemens Software Center to run malicious code through signed binaries. Affected organizations should immediately update the application to version 4.0.9.8.
7. A surge of active exploits targeting PHP, Node.js, and backdoored AI packages like LiteLLM caused widespread disruption across IT environments on October 21. The developments underscore the critical need for immediate patching and supply chain vigilance.
8. Shares of South Korean data center operator Elice dropped by nearly a third on its market debut due to rising borrowing costs and cooling investor hype. The fall reflects shifting financial dynamics around AI infrastructure funding.
9. Node.js released security patches across all supported versions to address a high-severity Brotli decompression vulnerability that could allow remote code execution. System administrators should upgrade Node.js installations promptly to mitigate compromise risks.
10. Sekoia observed threat actors targeting exposed Flowise AI orchestration servers to harvest stored credentials, inject malicious prompts, and hijack workflows. Organizations should avoid exposing AI agent management tools to the open internet.
11. Researchers in Tokyo reported that attackers are using China’s ARTEX AI penetration tool to target Japanese organizations. The campaign indicates expanded regional adoption of AI-driven attack platforms across East Asia.
12. LiteLLM disclosed that leaked GitHub tokens compromised its release pipeline, allowing attackers to publish a backdoored Python package to PyPI and Docker Hub. Developers using LiteLLM version 1.107.0 should remove the package immediately.
13. Ride-hailing platform Bolt announced plans to test autonomous robotaxis in Madrid using Lucid electric vehicles and Nvidia AI software. The trial marks an expansion of commercial self-driving technology in European urban markets.
14. Amazon revealed that threat actors attempted to exploit zero-day PHP2Shell vulnerabilities against honeypots four days prior to public disclosure. The finding confirms that attackers were actively probing for these web flaws ahead of patches.
15. China established a state-owned enterprise to centralize and manage national data assets, aiming to boost artificial intelligence development and digital economic growth. The move consolidates government supervision over critical data infrastructure.
16. Critical PHP vulnerabilities led to temporary outages across several European government websites after emergency patching was required. IT teams managing web applications must prioritize updating PHP stacks to ensure operational availability.
17. Law enforcement agencies disrupted nearly 11,000 cybercrime networks and arrested over 1,400 suspects across Africa during Operation Serengeti III. The operation focused heavily on business email compromise, ransomware, and online fraud.
18. State Bank of India confirmed a data breach after nearly 3,900 customer records appeared on dark web forums within a broader dataset leak. India's largest bank is actively investigating the origin of the exposed customer data.
19. A flaw in Signal's RingRTC library could permit call participants to continue receiving audio or video feeds after the call is terminated. Users should update Signal to receive the patched media library.
20. Attackers are actively exploiting a DLL side-loading bug in Siemens Software Center to deliver ClickFix malware via trusted application files. Siemens confirmed the flaw and released an automatic software fix.
21. An unidentified threat actor utilized the Chinese AI tool ARTEX to conduct automated cyberattacks against South Korean financial institutions. The incident underscores the rising risk AI tools pose to financial sector networks.
22. Cybercrime syndicate CyberCartel used commercial AI models, including Claude and Grok, to automate attack planning against over 400 financial institutions in Latin America. The activity shows how attackers leverage public AI services to scale campaigns.
23. German beverage company aquaRömer deployed G DATA’s Managed Extended Detection and Response service to protect its operational infrastructure and supply chain systems. The managed service assists the company's internal IT team with round-the-clock threat monitoring.
24. Threat actors are abusing genuine order emails from Dell and Staples to insert scam contact details and bypass email security filters. Recipients should verify unexpected order notifications directly through official corporate channels.
25. CISA added four actively exploited vulnerabilities—including the Plex2Shell flaws impacting Plex Media Server—to its Known Exploited Vulnerabilities catalog. U.S. federal agencies are required to apply patches within specified deadlines.

Summaries written with AI (Google Gemini) from the linked source articles.

Sources in this roundup

Most-mentioned keywords

Sources

1. How Some Malicious AI Skills Hide Instructions from Reviewers
2. Microsoft Links ConnectWise Hack To Chinese AI Tool
3. Chrome Adds New “Suspicious Site” Warnings — Here’s When You’ll See Them
4. Anthropic finds 94 cases where open-source projects drop security controls without warning
5. New QPACK Flaws Can Crash HTTP/3 Servers, Trigger Up to 4,000-Fold Memory Amplification
6. Hackers Exploit Siemens Software Center Vulnerability to Execute Malicious Code
7. October 21 Cybersecurity Nightmare: PHP and Node.js Flaws, LiteLLM Backdoor, and Active Attacks
8. Shares Of South Korean Data Centre Firm Plummet On Debut
9. Node.js Brotli Decoder Bug Could Allow Remote Code Execution in All Supported Versions
10. Attackers Target Flowise AI Agents With Credential Theft, Prompt Injection and Workflow Hijacking
11. Chinese AI Tool Used In Wave Of Japanese Hacks
12. LiteLLM Suspects GitHub Token Leak in AI Gateway Logs Behind Malicious Package Release
13. Bolt To Test Robotaxis In Madrid Next Year
14. Someone Tried to Exploit PHP2Shell 4 Days Before the Vulnerabilities Were Publicly Disclosed
15. Chinese Government Launches State-Owned Data Company
16. EU Website Outages Follow Critical PHP Flaws: What IT Teams Should Know
17. Authorities disrupt nearly 11,000 cybercrime networks in Africa
18. State Bank of India Confirms Data Leak After 3,897 Customer Records Found on Dark Web
19. Signal flaw could expose audio and video even after users hang up
20. Hackers Actively Exploiting Siemens Software Center Flaw in ClickFix Malware Attacks
21. Unknown Threat Actor Uses AI-Driven ARTEX to Target South Korean Finance
22. Hackers Used Claude and Grok to Automate Cyberattacks on Over 400 Banks and Financial Firms
23. aquaRömer relies on G DATA MXDR: “I’ve rarely seen such a smooth rollout”
24. Hackers Are Using Real Dell and Staples Emails to Make Their Scams Look Legitimate
25. CISA Orders Federal Agencies to Patch Exploited Plex2Shell, Siemens, and SiteGuard Flaws