I'm Tom Acklam. I spent decades building sales and marketing software for SMBs, and now I'm building SalesCitation. This page is for vCISOs and managed service providers with AI and data governance expertise that want their name to come up when an owner asks AI what to do about staff using AI tools on company data.
On October 28, 2026, Security Buzz reported on a survey called the Q3 2026 Pax8 SMB AI Pulse Report. Pax8 surveyed 1,150 business technology decision-makers in the United States, United Kingdom, Germany and Sweden. Among the small and midsize businesses working with AI, 35% had found unauthorized AI tools, workflows or agents running inside the company in the previous six months.[1]
I read the news report and have not read Pax8's own report. The article does not say how many of the 1,150 were in the United States or when they were asked. I use it because I haven't found a U.S. count of the same thing.
Employees are using AI tools nobody approved. How do I get found by owners looking for help with that?
Owners looking for help with unapproved AI can find you if your site says what to do in the days after they discover it. That owner is going to type something close to "someone here is using an AI tool we never approved, what should we do?"
Answer it in order. The Security Buzz article says oversight begins with identifying which AI systems are running, who is responsible for each one and what company data they can reach.[1] Start your page there, and say how you find them.
Then write about what staff build. The article says existing AI policies tend to cover which tools employees may use and say little about what employees build. Among the businesses in Pax8's survey that are building AI, 46% are building or testing agents that carry out multi-step tasks on their own.[1] A page about the agent somebody in accounting built over a weekend would answer a question the tool policies skip.
How do I get my vCISO practice recommended by ChatGPT?
A vCISO practice gives itself a chance with ChatGPT by publishing what it would tell an owner on a first call, under the name of the person who would take that call. No one can promise you the recommendation.
Check your own site first. Have whoever runs your site list the AI crawlers it blocks. OpenAI's crawler documentation says its settings are independent of each other: a site can allow OAI-SearchBot, which is for search, and disallow GPTBot, which collects content that may be used for training.[2] That lets you stay out of training without opting out of search.
Then run a test. Type a question the way an owner would: "We're a 45-person accounting firm. Staff have been pasting client data into ChatGPT. Who can write us a policy and put controls in place?" Read who gets named, then open the pages it cites.
Staff are picking up AI on their own with no training. How do I reach the owners who just figured that out?
Reach that owner by showing them how AI got into the business and what to do about it this month. Ipsos surveyed 1,070 adults employed at U.S. small businesses with 2 to 499 employees on May 8 to 11, 2026, for the U.S. Chamber of Commerce Foundation. Half said they use AI at work, and about one in 10 said they were offered formal AI training.[3]
The survey also asked who drove the adoption. About one in five workers, 19%, said it was driven mostly by employees exploring tools on their own, and 11% said it was driven by the organization's guidance or direction.[3] Those are small shares, and the article does not say what the rest answered. The Foundation says it is working with local chambers to bring AI training to small businesses, so it has a stake in these findings.
Another figure matters to a security firm. Nearly half of the workers, 47%, named privacy or security concerns as a barrier.[3] That tells me some of the staff are worried as well. Write for an owner who wants to give them rules they can follow. A one-page list of what must never be pasted into an AI tool is a first step.
Owners are worried about all the new state AI and privacy laws. How do I show up when they ask AI what applies to them?
Show up by answering for one state and one kind of business at a time. The U.S. Chamber of Commerce says in the 2026 edition of its small business technology report that six in ten small businesses worry about increased litigation and compliance costs from a patchwork of state privacy and AI laws.[4]
The Chamber is a business lobbying group, and its web page gives no sample size or survey dates. I take it as a sign the worry is real.
An owner does not ask AI about a patchwork. They ask whether one law applies to a company their size in their state. Write a page for each state you work in. Put the date you last checked it at the top. Tell them which questions are for a lawyer.
Do small businesses ask AI how to write an AI policy, and who does it send them to?
I haven't found a count of small businesses asking AI for an AI policy, and I can't tell you who it sends them to without testing it. The closest number I have is about workers. In the Ipsos survey, 88% of the workers who use AI and do research as part of their job use AI for research and information gathering.[3] That is employees at small businesses. It does not count owners writing policies.
It may be that AI drafts the policy itself and sends them to nobody. If that is right, the opening for you is the next question the owner asks: is this policy enough for a company like mine?
Test it before you believe me. Ask three AI tools to write an AI use policy for a 25-person company and note whether any firm is named or cited.
Should I publish a free AI use policy to get cited by AI?
I'd publish one. I can't promise it will be cited. I have not found a study that tests it.
In every country in Pax8's survey, fewer than half of the businesses had documented review and approval procedures for AI. The share ran from 23% in the United Kingdom to 45% in Sweden.[1] The article gives no figure for the United States.
Make yours cover what staff build as well as what they use. Who may build an agent? What data may it reach? Who owns it when the person who built it leaves? Put the policy in plain text on the page. If it is only a download behind a form, a crawler won't see it. Say what the template can't do for them.
Why does AI recommend software for AI security and not consultants?
I can't tell you why, and I would be wary of anyone who says they can. My guess is that software companies have published more pages about AI security than consultants have.
There is a sign that businesses want help from people. In Pax8's survey, seven in ten respondents said security concerns had led them to change their approach, slow adoption or seek outside help.[1] The article does not split that figure, so I can't say how many went looking for outside help.
Write about the decisions software leaves to a person. The article's advice is to limit each agent to the data and systems it needs and to monitor what it does after it goes live.[1] A product can do the monitoring. A person still has to decide what the agent may touch.
Sources
- Security Buzz, Small Businesses Face Growing Security Risks as They Build Their Own AI Tools (28 October 2026), reporting on the Q3 2026 Pax8 SMB AI Pulse Report
- OpenAI, Overview of OpenAI Crawlers (undated; read 17 November 2026)
- U.S. Chamber of Commerce Foundation, Half of Small Business Workers Use AI, Most to Boost Productivity, Not Automate Jobs (17 June 2026)
- U.S. Chamber of Commerce, Empowering Small Business: The Impact of Technology on U.S. Small Business (2 October 2026)