
When we see rapidly growing new technologies, it’s commonly delivered alongside talk tracks about the “everchanging cybersecurity threat landscape”, and we are also told to be vigilant. None of this rhetoric is particularly useful to someone who is primarily responsible for ‘doing the doing’ within an IT management setting; this also includes the management of cyber security products and their configuration.
This burden and constant stream of new things to worry about can make an IT technician feel overwhelmed. The list of things within the IT and cyber industry to manage only seems to get bigger, not least due to the new, sharp uptake in end-users vibe coding their own apps and expecting IT to figure out hosting, security, and support… But do not worry! While there is an engulfing amount of hysteria around all things AI, the protection of its use is not vastly different from what you already know.
Microsoft recently published their yearly Digital Defense Report, this report serves as my guiding priority list as to what I should be doing or advising and helping others with over the next twelve months.
Can a vendor-specific report apply to non-users of the products?
Yes, because in my experience, most of what I see in post-breach reports is the same pattern: product choice is not the main reason a business is compromised, it is how the products in use are configured. We see attacks dressed up as “sophisticated” when the root cause is someone misconfiguring a basic control that every product has.
In this year’s Digital Defense Report, Microsoft made a refreshing remark:
According to Microsoft Digital Defense Report, identity is the surface that matters most, and most of what we exploit there has nothing to do with AI. The core failure is the one red teams have exploited for years: too much standing access, too loosely enforced.
That should make anyone working in cyber security feel slightly more at ease. AI has accelerated the urgency to look after and close these gaps, but you already have the required experience to protect your customers.
What about AI Agents? Fundamentally they operate like virtual users. They have an identity, they access files and resources and carry out digital tasks using a computer. Because of that, identity must remain a priority.
Detecting and Responding to AI
Today it would not be wise to block all AI tools within your business, your userbase has become very dependent on them to achieve the results that they do today. Your first steps here are to discover what is being used.
This is likely most easily done by using tools that have access to the network traffic, most vendors offer something that gives you data on which domains and categories are being visited. When you present that to business owners, they might be shocked where the data is going. This exercise typically makes it clear for a company owner to understand that AI governance is not optional. Once you’ve got them onboard you can begin to survey how AI is used. Is it meeting notes, presentations, or responding to customer complaints? Or something more complex?
Technology is the quicker and easier bit of IT management. Understanding and working with people is what separates you from the rest, and it demands real effort. As quickly as I can download a new AI tool and start uploading business docs IT don’t know about, they can block it.
As the IT department, we know what is available and how to use it because we deployed it. But did the rest of the business get adequate training on the approved AI tools, and do they understand what not to do with business data? Are there pockets of the org who would be more productive by using more specialised AI tools? If the answer is yes, can access to those tools be routed through strong security controls?
What we should be asking people is “What is the business problem you are trying to solve?” The answer should not start with “I need AI to X…”. A business problem is not initially defined by the desired technology to solve it.
Once you have the needs defined, you can decide on the right tool, then begin to work out what these tools can see in the eyes of the end user. Now you’re straying into a data governance project, but that doesn’t make it the right place to start with your cybersecurity priority.
Securing the business in the correct order is what prevents most compromise.
What’s the bigger picture?
Up to this point, identity has held most of the attention in this article, but let’s look at the bigger picture. Few IT folk have been able to work in environments where they are involved with all aspects of the cyber security strategy, and when dropped into an MSP setting, it can be difficult to know where to begin. I think there is an order that people should follow:
- Identity Controls
- Email Protection
- Endpoint Protection
- Endpoint Management
- Files & Collaboration tool Management
- Data Governance
This path aligns with the most failures which lead to most business compromises, i.e. if you had limited time to make the biggest impact to prevent most breaches, you’d start at Identity and see how far you can get. However, all steps have their own merits, but some are more important than others. For instance, if you started with data governance but all users to access where that data is stored on any device, that effort has less impact than firstly preventing that action within the Identity project.
The priority remains ensuring that businesses implement the fundamental security controls that have long been considered essential. Getting these basics right creates a stronger security foundation and better prepares organizations to address the risks and challenges that come with emerging technologies.
About the author:
Lewis Barry is Principal Security Architect at inforcer. He is also a Microsoft Security MVP and Microsoft Business Premium advocate who takes a practical, cloud-first approach to IT and cyber security strategy. Read his blogs on Microsoft 365 Business Premium, Intune, MSP security, and more.
Join our LinkedIn group Information Security Community!












