October Retrospective

October in Review: 15 security topics, one common thread—the attackers didn’t break in; they used a key. Three themes and three things to do next month.

First topic: the stolen key. The FBI has issued a warning about the FortiBleed campaign: FortiGate firewalls exposed to the internet are being attacked using leaked passwords, and attackers are gaining administrator access and locking out the owners—approximately 86,000 devices have been compromised in nearly 200 countries. A fake “AI advertising tool” opens a login window overlaid on the page and steals Google, Meta, or TikTok advertising accounts, including two-factor authentication codes. ChatGPT accounts were targeted via a fake OpenAI invoice, and sponsored ads for “chatgpt” led to a fake GPT that installed remote-control software. In Romania, the DNSC warned about fake calls from “ONPCSB analysts” and fake text messages sent in the name of TPARK. None of these incidents required a security breach.

The second issue: access you didn’t authorize. Legitimate remote access programs, exploited by attackers, appear in nearly half of the incidents analyzed by Huntress—antivirus software doesn’t block them because they are genuine programs. The 15,465 public AI endpoints analyzed by OX Security are hosted on stores that don’t verify anyone, and 2.3% use expired domains that anyone can purchase. A spreadsheet can execute code upon opening in LibreOffice and OpenOffice if Java is enabled. And Copilot Business defaults to a pay-as-you-go model for new licenses purchased through a partner. These are all settings or access permissions that the provider or an application has chosen for you.

Third issue: The repaired site isn’t necessarily clean. cPanel had a vulnerability score of 9.4 out of 10, a WordPress backdoor is active in eight locations at once and self-replicates, and Ninja Forms and Bricksforge (10.0 out of 10) have been exploited in real-world attacks. A Google report shows why “we’ll do the updates later” no longer works: a vulnerability in a BeyondTrust product was exploited just four days after the fix was released. The update stops the attack, but it doesn’t clean up a site that’s already been compromised.

What's worth doing next month, in order of importance:

  • Two-factor authentication—preferably via passkey, email, VPN, and advertising or AI accounts—and no logging in, making payments, or installing software from a received link. It costs nothing and covers the first topic almost entirely.
  • A short list of access points: remote access programs, AI connectors, site administrators, and firewall administrators. Request this in writing from your IT and hosting providers; anything you don’t recognize should be removed.
  • Update the website and server quickly, then verify them: remove unknown administrators, third-party plugins, and keep a clean backup stored off-server.

NIS2 specifically requires access and vendor controls. You can check for free, in five minutes, whether your company falls under NIS2: xdn-cs.ro/instrument-nis2.html

Sources:

This article was generated with AI assistance.

Request a quote

← All news