Good morning. Three of today’s five stories come down to the same uncomfortable idea, which is that the thing letting an attacker in is not broken, it is working exactly as designed and being used by the wrong person. A firewall console, an employee’s phone, a marketing email platform. Here are the five stories that matter most if you are running a small or midsize business, or leading its IT.
Cisco’s firewall management software is under attack by three separate groups, including a ransomware crew and Russian state hackers. Cisco and CISA warned on Wednesday that CVE-2026-20079, a critical authentication bypass in Secure Firewall Management Center that Cisco patched back in early March, has been exploited since at least August, and CISA has given federal agencies until September 12 to fix it. Cisco’s Talos team is tracking three distinct activity clusters: one that dropped a web shell and a malicious JAR file to harvest credentials, one tied to the Russian APT known as Sandworm that delivered Cyclops Blink malware, and a third connected to the Qilin ransomware group that used a second FMC flaw to steal credentials and build a list of endpoints worth encrypting. That last detail is the one to sit with, because it describes an attacker doing target selection inside your network from the console that manages your firewall. This is the third FMC flaw added to CISA’s exploited list this year, and the pattern is familiar by now, since the fix existed for six months before anyone got hit. If you have an FMC appliance, two questions are worth asking today: is it patched, and is its management interface reachable from the internet, because Cisco says closing off that second one substantially reduces the risk on its own. Read more at SecurityWeek
Attackers are getting around Microsoft 365 security by calling employees on their personal phones. Microsoft researchers have tracked two initial access groups since May, Storm-3032 and Storm-3121, that phone or text staff on personal devices while posing as the company IT help desk, saying a passkey, MFA or single sign-on setting has to be updated immediately or work access will break. The link goes to a convincing Microsoft sign-in page where an adversary-in-the-middle setup or a device code flow captures the session, after which the attackers register their own MFA device for persistence and quietly query the Microsoft Graph API to inventory users, groups and sites before pulling small batches of files out of SharePoint, OneDrive and Exchange. Microsoft says that access is often handed off to extortion crews like ShinyHunters. The painful part for anyone trying to investigate afterward is that almost none of this touches corporate equipment, so the earliest evidence of the break-in is usually an employee remembering a phone call. Robert Coles of Black Duck makes the point that matters, which is that banning personal devices would not have stopped any of this, because the device was never compromised, the person was. What actually helps is phishing-resistant MFA on every sign-in, blocking device code authentication where you do not need it, and telling your team plainly that your help desk will never call them about a passkey. Read more at Dark Reading
Trezor’s customers were phished by an email that genuinely came from Trezor, because the break-in happened at the marketing platform in between. An attacker got into Brevo, the third-party email service Trezor uses for its newsletter, and used it to send mail from 120 customer accounts including Trezor’s. Recipients got a fake critical security alert from help@trezor.io warning that a hardware flaw in their wallet’s microcontroller could expose their recovery seed, with a link to an app that asked them to type in their wallet backup. Trezor took the domain down within twenty minutes, which held the damage to about 2,500 people who clicked, but the message reached roughly 347,000 newsletter addresses that are now a proven target list for the next attempt, and similar mail went out through BitBox and CoinTracking accounts on the same platform. This belongs on your desk even if you have never touched a crypto wallet, because almost every business now sends customer email through a service like this, and that service is authenticated to send as you. The questions worth answering this week are who has access to your sending account, whether it is protected by more than a password, and how quickly you could tell customers that mail went out over your name that you did not write. Read more at BleepingComputer
California has created the first state framework for independent AI audits, and what it signals matters more than what it requires today. Governor Newsom signed two bills on Wednesday: Senate Bill 813, which sets out how third-party verification organizations can assess AI systems and models for compliance with state law, and Assembly Bill 1405, which creates a state registry of AI auditors with standards for independence, transparency and integrity. The deadlines are years out, with the first class of verification organizations certified by January 2028 and the auditor registry live in 2029, so nothing lands on your calendar this quarter. Miranda Bogen of the Center for Democracy and Technology told CIO Dive she reads this as the first step toward a more complex ecosystem where audits become expected or required, and that for now it mainly helps companies know which assessors exist. The practical read for a smaller business is that “we use AI responsibly” is drifting away from being a statement you make and toward being a thing somebody else checks, particularly if your AI touches hiring, pricing or anything that materially affects a customer. Keeping a plain written record of what your AI tools do, what data they see and who reviews their output costs almost nothing now, and it is exactly the material an assessor or a customer’s procurement team will ask for later. Read more at CIO Dive
A security services chief executive says not one security leader he has spoken with has been given extra budget to secure AI. Chris Schueler, CEO of Cyderes, said in a recent podcast that every CISO he has talked to is being asked to protect AI and agentic tools out of the budget they already had, even while their companies spend heavily on the AI itself. He expects a genuine renaissance in security operations from AI over the next two years, and he argues the responsible approach is to carve a portion of every significant AI investment into the controls and guardrails that go with it, rather than waiting for a large AI-related breach to unlock the money. That framing translates cleanly to a company of any size, and it is worth checking against your own numbers, because budgets tend to tell the truth about priorities. If you approved a new AI tool this year and nothing in your security or compliance line moved, the gap Schueler is describing already exists in your business. The simple version of the fix is to treat the guardrails as part of the price of the tool at the moment you approve it, the same way you would not put a delivery van on the road without insuring it. Read more at CRN
Follow Up
Follow-up to the August 31 PaperCut story: researchers now say one attacker ran the entire campaign with hundreds of AI agents and reached 395 organizations. GreyNoise traced the operation, which began August 31 against two PaperCut NG/MF flaws, and found the attacker used hundreds of agents built on OpenAI’s Codex and DeepSeek models, plus the Netlas scanning service for target lists, to build, test and refine the exploits. The numbers are what earn it a second mention: at least 440 compromised PaperCut instances tied to 395 organizations across 48 countries, credentials harvested from 280 victims, operating system or domain secrets from 147, and administrator rights at 12. The attacker went from an empty workspace to remote code execution against a real victim in under four hours, reached domain admin two hours after that, and once the campaign launched, compromised 11 organizations in 26 seconds. The lesson is not that AI invented a new attack, since these were known flaws with patches available, but that the window between a patch going public and your unpatched server being found is now measured in hours rather than weeks. Read more at The Register
Sponsored by Lucky 13 Solutions
Business in Motion. Tech in Sync. Lucky 13 Solutions is a managed services provider helping small and midsize businesses keep their IT reliable, secure, and well-supported, without needing a full in-house team. Learn more at l13s.com.
If you do one thing with this edition, find out whether anything that manages your firewall is reachable from the internet.
Get the Business IT News Roundup in your inbox:
How this gets made: stories come from my own reading and from AI-assisted research and drafting, all reviewed and edited by me before anything goes out. The commentary is mine, sources are linked so you can read the original, and any errors are mine to own. If something looks wrong, reply and tell me.