How AI is changing the risks facing small businesses
Global website vulnerabilities, freezers being sent into meltdown across the continent, and warnings that the UK faces an attack on its critical infrastructure that could wipe out food supplies and the electric grid in two years. It seems that barely a day goes by without a new story about AI going rogue or the potential of cyber hacks to undermine our daily lives. John Speed, the founder of IT solution provider Heliocentrix, argues that now is the time for SMEs to adopt a wider approach to digital safety.
The changing environment of AI threats
The recent urgent warning about vulnerabilities in PHP, a technology behind many websites, brought a wider change into focus for us. We found ourselves waiting for a security update to be released so we could assess it and start updating affected client systems. Against the UK’s National Cyber Security Centre’s (NCSC) warning that AI could accelerate the discovery and exploitation of software weaknesses, it was a practical reminder that the pace of technology risk is changing.
For SME owners, that change has to compete with customers, staff, cash flow, regulation and growth. Many have managed technology themselves, drawing on personal experience, advice and common sense, then calling someone when things became complicated. For many owners, that has felt like a workable approach.
The problem isn’t that business owners have suddenly become careless. The environment around them is changing. AI is changing the threats coming from outside the business and, as we give it more access and authority, creating new risks inside it. The PHP vulnerabilities highlighted the urgency of the first; unexpected AI-related bills show why the second deserves attention too.
Why are there so many alarming headlines about AI?
AI inevitably attracts sensational headlines. Stories about systems behaving unexpectedly, hackers using AI or autonomous agents doing things their users hadn’t anticipated are more interesting than stories about technology working normally.
The NCSC has warned organisations to prepare for a ‘vulnerability patch wave’. Its concern is that AI is showing an ability to exploit years of accumulated weaknesses in software ‘at scale and at pace’. The NCSC expects this to result in more vulnerabilities being identified and a rush of updates across different types of software.
That matters to SMEs because the economics of cybercrime are changing. As AI makes parts of finding and exploiting weaknesses faster and easier, attackers can potentially target more organisations with fewer resources.
The PHP warning showed what AI vulnerabilities can look like
We saw a particularly good example in October.
On 21 October 2026, the NCSC issued an alert about two critical vulnerabilities affecting PHP, a technology widely used behind websites and web applications. The vulnerabilities created the possibility that criminals could take control of affected websites and systems remotely.
The advice was unusually stark. Organisations were told to identify where they were running PHP, prioritise internet-facing systems and install the security updates. Those unable to update or apply temporary mitigations were advised to “consider taking the system offline until it can be fully mitigated.”
Software updates have traditionally been easy to view as routine maintenance. Something the person responsible for the website or IT can get around to when there is time. We are increasingly entering an environment where the gap between “there’s an update available” and “we need to act” could be much shorter.
We experienced that first-hand. A number of servers we manage for clients relied on technology affected by the vulnerability. We knew a security update was due for release at 3pm, so we were sitting ready for it to arrive so that we could assess it and start updating affected systems.
That isn’t a situation we’ve routinely encountered before. Updates are part of everyday IT management, but having a major vulnerability, unusually stark NCSC guidance and professionals waiting for a specific security release felt different.
Why should SMEs be concerned about AI risk?
An attacker doesn’t necessarily choose a particular 25-person business and spend weeks trying to break into it. Internet-facing systems can be discovered and tested automatically. If thousands of organisations are running vulnerable technology, attackers can look for opportunities across them at scale.
A larger organisation might have an internal security team and established processes for monitoring vulnerabilities. An SME can be running exactly the same vulnerable software without anyone whose working day includes checking whether an urgent security advisory has just been published.
The internal risks to SMEs of AI adoption
AI isn’t only changing the threats coming towards your business. It is also changing what happens inside it, as we give AI tools access to information, permission to act and, sometimes, the ability to spend money.
Some AI tools go beyond answering questions. AI agents can carry out tasks in connected systems, using the access we give them. That can deliver significant benefits, but it changes the consequences when something goes wrong.
Imagine giving a new employee access to company systems, customer information and a credit card, then asking them to complete a task without explaining what they can spend or which decisions need approval.
Few business owners would consider that sensible. Yet we can inadvertently give an AI agent the digital equivalent of those permissions.
One reported incident involved an AI agent given access to Amazon Web Services (AWS). While attempting a networking task, it created expensive cloud systems and repeatedly duplicated them. The resulting charge reached about $6.5k before its operator noticed, although AWS later reduced the bill.
Larger unexpected AI-related bills have also been reported, but not all have the same cause. In separate cases involving stolen access credentials, criminals used customers’ cloud AI services and left them facing charges of tens of thousands of dollars.
Whether the cause is an AI doing more than expected, an employee unknowingly consuming expensive services or criminals stealing credentials, the lesson is similar: if technology has permission to spend your money, that permission needs controls.
Should SMEs be worried about using AI?
Worried isn’t particularly useful. Careful is.
AI can help smaller businesses achieve things that previously needed considerably more time, people or specialist skills. Avoiding it altogether because there are risks makes little sense. But businesses shouldn’t confuse something being easy to use with it being risk-free.
If AI can see sensitive company information, consider what it actually needs. If it can take actions, decide which require human approval. And if it can purchase services or use cloud services charged according to consumption, make sure appropriate limits and monitoring exist. The greater the authority you give an AI system, the more important those controls become.
Think about digital safety, not just cybersecurity
Traditional cybersecurity remains essential. Businesses still need supported software, timely updates, multi-factor authentication, sensible access controls and reliable backups. But those measures are becoming part of a wider issue: digital safety.
That means protecting the business from external threats while keeping the technology used internally under control. AI makes both sides more important: it is changing what attackers can do and what we allow our own systems to do.
Start with visibility. Know which technology your business depends upon and who is responsible for it. The NCSC specifically recommends prioritising systems exposed to the internet when dealing with vulnerabilities.
This isn’t about adding another layer of fear or expecting SME owners to become cybersecurity and AI experts. It’s about recognising when technology needs proper ownership, controls and professional advice. Somebody must own both the response to external threats and the controls over AI used inside the business.
If you’re looking for office space from a provider that provides free expert advice to help you succeed, get in touch.
MEET THE EXPERT

John Speed is the founder and director of Heliocentrix, which offers IT Services to SMEs throughout the United Kingdom, creating bespoke IT solutions tailored for each organisation’s strategy, structure and people and designed to meet both current and future needs. Their IT solutions are from world-class technology providers, implemented and supported by their expert team.