Daily cybersecurity briefing

Top 10 Cybersecurity Stories for October 22, 2026

PHP released fixes for PHP2Shell’s two unauthenticated code-execution flaws after Cloudflare reported early attacks; administrators should deploy PHP 8.5.12, 8.4.27, 8.3.36, or 8.2.35, since WAF rules and configuration mitigations do not fully block exploitation. Patchstack reported that an attacker used valid UpdraftCentral dashboard keys to issue commands that installed backdoor plugins on connected WordPress sites.

Compiled by the Slugnet Editorial System. Published Oct 22, 2026, 8:03 AM EDT

Audio briefing

Listen to this edition

A spoken version of today’s prioritized cybersecurity briefing.

  1. 01
    BleepingComputer

    PHP ships emergency patches for critical PHP2Shell flaws

    What happened

    PHP released emergency fixes for CVE-2026-109618 and CVE-2026-109619, collectively called PHP2Shell, two unauthenticated remote code execution flaws affecting PHP installations that serve web requests. Both involve corrupted memory state after abnormal request termination and can allow code execution with web worker privileges; Cloudflare reported observing early attacks before patch release. Administrators should install PHP 8.5.12, 8.4.27, 8.3.36, or 8.2.35, or obtain vendor-backed fixes for older branches; PHP developers warn that temporary mitigations do not fully block exploitation.

    Why it ranks #1

    PHP released emergency fixes for two unauthenticated remote code execution flaws affecting web-serving installations; Cloudflare reported pre-patch attacks, and PHP developers warn temporary mitigations do not fully block exploitation.

    Who should care

    Application security teams, IT and platform operations, SOC and incident response teams

    What to do

    Install PHP 8.5.12, 8.4.27, 8.3.36, or 8.2.35 as soon as possible on servers running PHP 8.x, and migrate servers running PHP 7 or earlier to a fixed release.

    Impact
    high
    Urgency
    immediate
    Confidence
    high
    Scope
    product
    Status
    disclosed
    Read the original source Link to this ranking Share on Bluesky Share by email
  2. 02
    BleepingComputer

    Japan's Onamae Desktop Cloud hacked, up to 27,000 servers affected

    What happened

    Onamae Desktop Cloud, operated by GMO Internet, confirmed unauthorized access and malware installation in some customer environments, with up to 26,739 customer servers potentially affected. The provider began shutting down customer servers at 04:00 local time on October 21, 2026, affecting all Desktop Cloud plans, and provided no restart estimate. Onamae Desktop Cloud confirmed that an unauthorized third party installed malicious programs in some customer server environments and warned that stored data may have been exposed, but said it had not confirmed any leakage or misuse and that payment information was unaffected because it was stored on separate infrastructure.

    Why it ranks #2

    Onamae Desktop Cloud confirmed malware in customer environments and began shutting down servers across all plans, with 26,739 servers potentially affected and no restart estimate.

    Who should care

    Cloud security teams, IT and platform operations, SOC and incident response teams

    What to do

    Rotate secrets, such as passwords, stored on your servers as soon as possible, as they may have been compromised.

    Impact
    critical
    Urgency
    immediate
    Confidence
    high
    Scope
    enterprise
    Status
    confirmed incident
    Read the original source Link to this ranking Share on Bluesky Share by email
  3. 03
    The Hacker News

    LiteLLM Suspects GitHub Token Leak in AI Gateway Logs Behind Malicious Package Release

    What happened

    Attackers compromised LiteLLM’s release pipeline on October 21, publishing version 1.107.0 with a credential stealer and an hourly remote-code loader to PyPI and container registries. They used developer GitHub credentials to alter protections and run a workflow against a lookalike repository, producing signed malicious artifacts; LiteLLM suspects the tokens came from response logs in its exposed internal AI gateway, but that access path remains unconfirmed. Wiz reported subsequent use of stolen AWS credentials to validate access and enumerate cloud resources, beginning roughly 4.5 hours after exfiltration.

    Why it ranks #3

    Attackers compromised LiteLLM’s release pipeline on October 21 to publish version 1.107.0 with a credential stealer and an hourly remote-code loader to PyPI and container registries, and Wiz reported that use of stolen AWS credentials to validate access and enumerate cloud resources began roughly 4.5 hours after exfiltration.

    Who should care

    Cloud security teams, IT and platform operations, SOC and incident response teams

    What to do

    Remove LiteLLM 1.107.0 from your machines, isolate hosts where the malicious version was installed, hunt for malicious Python artifacts, and block access to the exfiltration domain.

    Impact
    high
    Urgency
    near-term
    Confidence
    high
    Scope
    enterprise
    Status
    confirmed incident
    Read the original source Link to this ranking Share on Bluesky Share by email
  4. 04
    BleepingComputer

    ClickFix has never been bigger, and AI tools are its most convincing lure yet

    What happened

    Push Security reported AI-themed ClickFix campaigns using cloned installer pages and shared chatbot content to persuade users to execute malicious commands. In one fake Claude installer, the page displayed Anthropic’s genuine command while its copy button substituted a command that downloaded and executed an attacker-controlled script; a separate custom GPT directed users to a Google Sites page serving a Windows ClickFix payload. Microsoft separately reported that Defender observed ClickFix-style attacker-supplied commands executing on more than 1.1 million unique devices between February and early May 2026.

    Why it ranks #4

    Push Security reported AI-themed ClickFix campaigns designed to persuade users to execute malicious commands, including a fake Claude installer page that displayed Anthropic’s genuine installation command but used its copy button to substitute a command that downloaded and executed an attacker-controlled script.

    Who should care

    CISOs and security leaders

    Impact
    critical
    Urgency
    monitor
    Confidence
    high
    Scope
    enterprise
    Status
    emerging
    Read the original source Link to this ranking Share on Bluesky Share by email
  5. 05
    The Hacker News

    China-Aligned Bronze Sandcastle Abuses Sandboxie DLL Sideloading to Deploy Backdoors

    What happened

    Sophos observed legitimate Sandboxie executables alongside a malicious sbiedll.dll in multiple victim environments, where SandboxieWUAU.exe loaded attacker-controlled code during execution. The attackers placed a malicious sbiedll.dll alongside legitimate Sandboxie executables, causing SandboxieWUAU.exe to load an encrypted config.ini payload into memory; the backdoor can enumerate networks and accounts, log keystrokes, and capture screenshots. Sophos recommends monitoring for DLLs that read local files and immediately execute code in memory, because runtime decryption can limit static file-based detection.

    Why it ranks #5

    Sophos observed Sandboxie DLL sideloading in multiple victim environments, loading an encrypted backdoor into memory with network enumeration, keystroke logging, and screenshot capture capabilities.

    Who should care

    Application security teams, IT and platform operations, SOC and incident response teams

    What to do

    Focus instead on behavioral indicators such as DLLs reading local files followed by immediate in-memory execution, which may indicate decoding and staging activity.

    Impact
    high
    Urgency
    near-term
    Confidence
    high
    Scope
    enterprise
    Status
    emerging
    Read the original source Link to this ranking Share on Bluesky Share by email
  6. 06
    The Hacker News

    China-Linked Hackers Use ARTEX AI Agents in ConnectWise Supply Chain Attack

    What happened

    Microsoft attributed the ConnectWise Cloud Backup supply-chain compromise to Storm-0940, a group it assesses as China state-aligned, and said the attackers used the ARTEX agentic AI framework to automate access to customer environments. Beginning October 15, the attackers used stolen ConnectWise application credentials and existing tenant permissions to authenticate without interactive user sign-ins; Microsoft assessed that some activity progressed to reconnaissance, credential changes for persistence, and changes to identities, groups, and privileged roles. ConnectWise temporarily shut down services and is restoring them in phases.

    Why it ranks #6

    Stolen ConnectWise application credentials enabled access to customer environments without interactive sign-ins; Microsoft assessed that some activity progressed to persistence and privileged-role changes, while ConnectWise temporarily suspended services.

    Who should care

    Application security teams, Cloud security teams, IT and platform operations, SOC and incident response teams

    Impact
    high
    Urgency
    near-term
    Confidence
    high
    Scope
    enterprise
    Status
    confirmed incident
    Read the original source Link to this ranking Share on Bluesky Share by email
  7. 07
    The Hacker News

    New QPACK Flaws Can Crash HTTP/3 Servers, Trigger Up to 4,000-Fold Memory Amplification

    What happened

    FoxIO disclosed QPACK header-decompression flaws in seven HTTP/3 stacks and released proof-of-concept code showing how an unauthenticated attacker can exhaust server memory or crash servers through a single connection. Affected implementations fail to limit decompressed headers adequately; FoxIO measured up to 4,000-fold memory amplification in Bun, where a roughly 516 KB stream also triggered an unrecoverable CPU loop. Fixes are available for quic-go, Cloudflare quiche, lsquic and Wireshark, while Bun, the h3 Rust library and aioquic remained unpatched at disclosure; no attacks in the wild, data theft or code execution were reported.

    Why it ranks #7

    FoxIO disclosed QPACK flaws across seven HTTP/3 stacks with proof-of-concept code for unauthenticated, single-connection memory exhaustion or crashes; three implementations remained unpatched at disclosure.

    Who should care

    CISOs and security leaders, SOC and incident response teams

    Impact
    high
    Urgency
    monitor
    Confidence
    high
    Scope
    enterprise
    Status
    proof of concept
    Read the original source Link to this ranking Share on Bluesky Share by email
  8. 08
    The Hacker News

    Attackers Hijacked Over 300 Tor Exit Relays to Intercept SSH Connections

    What happened

    Researcher kaba found 333 Tor exit relays redirecting SSH traffic to 12 interception servers on October 2; passwords could be stolen if users accepted the interceptor’s host key and authenticated with a password. Attackers changed relay configurations to permit SSH traffic, and a honeypot captured follow-on attempts to use stolen credentials and install a backdoor. By October 5, affected relays had been removed from Tor’s consensus; the victim count and how most relay hosts were compromised remain unknown.

    Why it ranks #8

    Attackers redirected SSH traffic through 333 Tor exit relays, with a honeypot capturing stolen-credential reuse and backdoor installation attempts; the affected relays were removed from Tor’s consensus by October 5.

    Who should care

    CISOs and security leaders, Identity and access teams, IT and platform operations, SOC and incident response teams

    Impact
    high
    Urgency
    near-term
    Confidence
    high
    Scope
    enterprise
    Status
    confirmed incident
    Read the original source Link to this ranking Share on Bluesky Share by email
  9. 09
    The Hacker News

    Attackers Abuse UpdraftCentral Keys to Plant Backdoors on WordPress Sites

    What happened

    Attackers used valid UpdraftCentral dashboard keys to install backdoor plugins on connected WordPress sites on October 18–19, 2026, according to Patchstack, following a breach of TeamUpdraft’s website that may have exposed customer connection keys. Signed commands passed through the service’s remote-management channel to WordPress’s standard plugin installer; the backdoors could create privileged accounts that persisted after plugin removal. TeamUpdraft released UpdraftPlus 1.26.13 and Premium 2.26.13 to invalidate older UpdraftCentral connections and advised affected customers to inspect plugins and accounts and reset credentials.

    Why it ranks #9

    Attackers used valid UpdraftCentral keys to install WordPress backdoors capable of creating privileged accounts that persisted after plugin removal, making plugin cleanup alone insufficient.

    Who should care

    Identity and access teams, IT and platform operations, SOC and incident response teams

    Impact
    high
    Urgency
    near-term
    Confidence
    high
    Scope
    enterprise
    Status
    confirmed incident
    Read the original source Link to this ranking Share on Bluesky Share by email
  10. 10
    The Hacker News

    Pegasus Spyware Infection Confirmed on Serbian University Rector's iPhone Days Before Election

    What happened

    Citizen Lab said on October 19 that forensic analysis confirmed Pegasus spyware infected the iPhone of University of Belgrade rector Vladan Đokić, a supporter of Serbia’s student protest movement, through a zero-click exploit in 2026. The disclosure came six days before Serbia’s snap parliamentary election, but the lab did not identify the operator, specify the infection date or disclose what data, if any, was taken. Citizen Lab’s investigation remains ongoing, and its note does not identify the flaw used to infect Đokić’s iPhone or state whether current iOS releases block it.

    Why it ranks #10

    Citizen Lab said on October 19 that forensic analysis confirmed Pegasus infected University of Belgrade rector Vladan Đokić’s iPhone through a zero-click exploit in 2026, but it did not identify the operator, specify the infection date or flaw used, disclose whether any data was taken, or state whether current iOS releases block the exploit.

    Who should care

    IT and platform operations, SOC and incident response teams

    Impact
    high
    Urgency
    monitor
    Confidence
    high
    Scope
    enterprise
    Status
    emerging
    Read the original source Link to this ranking Share on Bluesky Share by email