top of page

Is Your IT Person a Cowboy? 7 Questions Every Owner Should Ask

3 minutes ago
7 min read

The video uses a digital avatar of Erick Grau, CEO of Chibitek.

Fair warning: this one is a bit of a rant. Over the last few weeks, some of the biggest names in tech and government had a rough stretch, and the reasons were not exotic. They were basics. I want to walk through what happened, what it means for a business like yours, and the seven questions I think every owner should be asking whoever runs their IT.

A rough few weeks for the big names

The FBI was reportedly hacked. ABC News reported that a law enforcement source confirmed a national crime database was hacked, and NBC News reported the FBI is investigating an apparent breach. On Fox News Sunday, the Director of National Intelligence said the FBI is looking into it. Nobody has said publicly what, if anything, was taken, and the FBI itself has not confirmed details. Several outlets describe it as the second reported breach affecting the FBI in recent months. I'm not going to guess at who or how. The point for us is simpler: even an agency with a budget request of $10.1 billion for fiscal 2026 is dealing with this.

Microsoft 365 email went down around the world. On November 12, most users could not get into their mailboxes through Outlook. Microsoft's own incident report says the root cause was a defect introduced months earlier, in a fix that went out in July, and that it sat quietly until the right load conditions hit. In its final report, Microsoft wrote: "Our testing did not adequately represent the combination of hardware and traffic mix encountered during this incident." Microsoft also said it was not a cyberattack.

The FAA called a nationwide ground stop. According to the FAA, its legacy telecom network was flooded at a rate of 1,000 connection attempts every 4 to 5 seconds after its network contractor deployed new routers at two sites. The FAA says it has no indication this was a cyberattack, and it paused installing those routers. This is an agency whose fiscal 2026 request was $22 billion, including $1 billion to modernize its telecommunications.

PHP servers came under attack. Last month CISA added two PHP flaws to its Known Exploited Vulnerabilities catalog and told organizations to patch right away. One security firm documented a single operator using a farm of AI agents to break into unpatched PHP servers within days of the flaws going public.

And the basics still bite: a bug that slipped past testing, a new router that flooded a network, servers nobody patched

None of this was a movie plot. A bug that slipped past testing. A network change that flooded a system. Servers nobody patched. If it can happen to organizations that spend billions, it can happen to a 40 person company with one IT person.

And then the banks changed the rules

This is the one I want owners to read twice. In the last few weeks, several big banks rewrote their online banking terms around AI tools that log in on your behalf.

  • Citi added a section on third party AI agents. If you give an AI agent your online banking login, Citi says it will consider the agent's access and actions as authorized by you. If those credentials are "further shared, leaked, or otherwise distributed," Citi will treat that access as yours too.

  • Wells Fargo's updated online access agreement, effective Tuesday, November 24, covers Wells Fargo Online and Business Online. It says transactions an AI agent starts are deemed initiated and authorized directly by you, and that "we do not monitor, screen, or review AI Agent activities."

  • Chase says outside AI tools may not use your login credentials, and allows access only through approved connections with your authorization.

  • PNC's digital services agreement, effective December 19, tells customers not to give their user ID, password or one time passcode to a third party AI agent.

Citi, Chase and Wells Fargo rewrote their online banking terms: your login plus an outside AI tool counts as you

The terms include language like "unless otherwise provided by law," and how consumer protection rules apply here is still being worked out. I'm not a lawyer and this isn't legal advice. But the direction is clear. If someone on your team hooks an AI tool up to the company bank login, the bank may treat whatever that tool does as you doing it.

Is your IT person a cowboy?

This next part is my opinion. A lot of small and mid sized businesses run on one IT guy, or a small shop that just doesn't take security seriously enough. Because it's not their company. Some are great. I mean that. But some of them are cowboys.

Here's what a cowboy setup looks like when I walk in:

Cowboy IT red flags: passwords in their head or on a sticky note, a backdoor nobody else knows about, no MFA on their own account, backups that should be fine, new apps with no sign-off, an AI tool hooked up to the bank login, no paper trail
  • Admin passwords that live in one person's head, or on a sticky note

  • A backdoor or two that nobody else knows about

  • No MFA on their own admin account

  • Backups that "should be fine," with nobody able to say when they were last tested

  • New apps installed with nobody's sign-off

  • An AI tool hooked up to the company bank login

  • No paper trail for any of it

The numbers back this up

I don't want you to take my word for it, so here's what the big annual reports say.

IBM's 2025 Cost of a Data Breach report found that one in five organizations reported a breach due to shadow AI, meaning AI tools used without IT approval. Organizations with high levels of shadow AI saw higher breach costs than those with little or none.

Verizon's 2025 Data Breach Investigations Report found that credential abuse (22%) and exploitation of vulnerabilities (20%) were the leading ways attackers got in. Third party involvement in breaches doubled to 30%. And for small and mid sized businesses, ransomware was present in 88% of breaches.

Verizon 2025 Data Breach Investigations Report: 30% of breaches involved a third party, double the year before; 88% of small business breaches had ransomware

Stolen logins, unpatched systems, outside parties and ransomware. That's the same list of basics.

7 questions to ask your IT person

These are the questions I'd want answered if I were in your seat. Ask them politely, and ask for the answers in writing.

7 questions for your IT person: get the answers in writing
  1. Who has admin access, and do they all still need it? Are there any backdoors? Every admin account should be documented and accounted for.

  2. Is MFA on every admin account and every remote login? No exceptions. A Microsoft study of commercial accounts found MFA reduced the risk of compromise by 99.22%.

  3. When did we last actually restore from backup? Not "the backup ran." A real restore. And is there a copy kept offline, where ransomware can't reach it?

  4. Do we have a list of every device, app and cloud account? And does any app or AI tool have our bank login? You can't protect what you don't know you have.

  5. What isn't patched right now, and when will it be? There should be a real answer with dates, not a shrug.

  6. Where's our written incident plan? Is there a printed copy, and do we know who to call first? If the network is down, a plan that only lives on the network doesn't help.

  7. Who checks your work? Everyone needs a second set of eyes. That includes me.

Good IT people welcome these questions. If yours gets defensive, that tells you something.

A note to fellow MSPs

I appreciate the good MSPs out there, and there are a lot of you. This applies to us too. Joint guidance from CISA and partner agencies says MSPs should put MFA on every account that has access to customer environments and treat those accounts as privileged. The same guidance calls for offline backups and printed copies of incident plans. If we're asking owners to hold their IT accountable, we should be the first ones to hand over the answers.

Key points

  • The recent failures at big organizations came down to basics: testing, change control, patching and access.

  • Several big banks now treat actions taken by an AI tool with your login as your own.

  • Shadow AI, stolen logins and unpatched systems show up again and again in breach reports.

  • Seven questions, answered in writing, will tell you a lot about how your IT is really run.

FAQ

Is my IT person a cowboy if they can't answer all seven questions today? Not necessarily. What matters is whether they take the questions seriously and come back with real answers and a plan.

Should we stop using AI tools? No. Use them deliberately. Get IT's approval, keep a list of what's connected to what, and don't hand an AI tool your bank login without reading your bank's terms first.

What's the single most important question? If I had to pick one, MFA on every admin account and remote login. But the one that keeps everything else honest is number seven: who checks your work?

Bottom line

Hold your IT accountable. Not because they're bad people, but because it's your company, your data and your money. If you want a second set of eyes on your answers, give me a call or email us at hello@chibitek.com.

Chibitek. Let us do IT so you don't have to.

Erick Grau, CEO of Chibitek

Sources

Comments


bottom of page