# Archive

Monday, October 26, 2026

1. 1

   1

   The Hacker News general

   PHP2Shell Exploitation Hits Government Entity Amid Global Scanning and Signs of AI-Assisted Hacking

   Threat actors exploiting PHP2Shell vulnerabilities (CVE-2026-109618 and CVE-2026-109619) have deployed agentic AI platforms for post-exploitation, compromising a European government entity within two hours of initial access to extract password hashes and MFA data. The use of AI-assisted post-exploitation marks a significant escalation in attacker sophistication that security teams need to account for in detection and response playbooks.

2. 2

   0

   BleepingComputer general

   PHP2Shell security flaws now sucessfully exploited in the wild

   Two critical RCE vulnerabilities tracked as PHP2Shell (CVE-2026-109618 and CVE-2026-109619) are being actively exploited in the wild, representing an urgent patching priority for any organization running affected PHP installations. The in-the-wild exploitation, combined with confirmed government-sector compromises, makes this a high-priority threat for defenders.

3. 3

   0

   The Hacker News general

   Microsoft 365 Token-Replay Attacks Compromise 97 Accounts at 79 Organizations

   Huntress researchers documented an adversary-in-the-middle token-replay campaign against Microsoft 365 that ran from July 14 to September 22, 2026, capturing sign-in sessions and replaying stolen tokens to compromise 97 accounts across 79 organizations, with 497 successful sign-in events detected. This campaign demonstrates that MFA alone does not protect against AitM token theft, reinforcing the need for conditional access policies and session anomaly detection.

4. 4

   0

   BleepingComputer general

   BlackX UEFI bootkit exploits Baton Drop flaw to bypass Secure Boot

   A UEFI bootkit dubbed BlackX, deployed in an espionage campaign in early 2026, leverages the known Baton Drop Secure Boot bypass vulnerability to load a malicious driver specifically targeting security software from Kaspersky and Avast. The bootkit's ability to survive OS reinstalls and actively subvert endpoint security products makes it a severe persistent threat requiring firmware-level detection capabilities.

5. 5

   0

   SecurityWeek general

   Node2Shell: Brotli Flaw Triggers Patching Across Microsoft, Node.js, Python Packages

   An out-of-bounds write vulnerability in Google's Brotli decoder, dubbed Node2Shell, affects a broad ecosystem including Node.js, Microsoft .NET, PowerShell, and Python packages, and is exploitable for remote code execution. The cross-ecosystem scope of this flaw demands urgent patching across multiple language runtimes and Microsoft products.

6. 6

   0

   The Hacker News general

   Gheist Uses Stolen GitHub Tokens to Steal Private Repository Secrets via Actions Logs

   A threat actor tracked as Gheist is running an automated campaign using stolen GitHub tokens to inject malicious workflows that exfiltrate repository secrets via GitHub Actions logs, then delete the run evidence — successfully targeting both private repositories and GitHub Enterprise environments, per Wiz's analysis. Security teams using GitHub Actions should audit token permissions and monitor for unauthorized workflow modifications or unexpected Actions log downloads.

7. 7

   0

   The Record threat-intel

   Popular AI tool LiteLLM says unrevoked key allowed hackers to target users

   LiteLLM, a widely-used open-source AI API proxy, confirmed that an API key accidentally exposed in May 2026 was leveraged in a supply-chain attack believed to have affected tens of thousands of users the week of October 19, 2026. The incident underscores the risk of long-lived unrevoked credentials in open-source AI tooling with large developer adoption.

8. 8

   0

   Schneier on Security threat-intel

   Bias in Qwen3-Coder Model

   Research on Alibaba's Qwen3-Coder model reveals that a single-word prompt change — specifying code is 'for a US defense contractor' versus 'for a Chinese defense contractor' — makes the model 19% more likely to produce insecure code, with the bias traced to CCP alignment embedded in the model's weights. This finding has direct supply-chain implications for Western enterprises integrating Chinese-origin AI models into development pipelines.

9. 9

   0

   The Hacker News general

   Malicious File in Downloads Folder Tries to Trick AI Coding Agent Into Installing Malware

   Security firm Forever Security caught a real-world prompt injection attack in which a malicious AGENTS.md instruction file in a developer's Downloads folder attempted to direct an AI coding agent to install a known-malicious npm package, stopping the agent before execution. The incident represents a concrete example of indirect prompt injection targeting AI coding agents and illustrates new attack surface considerations for developer security.

10. 10

    0

    The Hacker News general

    Unpatched Tailscale SSH Flaws Could Let a Hacked Device Reach Root on Other Servers

    A researcher disclosed that a single compromised device on a Tailscale network can be leveraged to gain root access on other servers reachable by that device, provided Tailscale SSH is configured to allow port forwarding and the attacker controls a device permitted by the network's ACLs. Organizations using Tailscale SSH for internal server access should audit port-forwarding configurations and review ACL policies to limit lateral movement potential.