Artificial intelligence is changing cyber security on both sides. Attackers use it to work faster and fool people more easily; defenders use it to spot attacks sooner. For a small business, or an MSP looking after dozens of them, the useful question is not "how clever is AI?" but "what should we do differently on Monday morning?"
What the NCSC actually said
In its assessment The near-term impact of AI on the cyber threat, the UK's National Cyber Security Centre judged that AI will almost certainly increase the volume and heighten the impact of cyber attacks. Importantly, it expects this mainly through the evolution and enhancement of existing tactics, not through entirely new kinds of attack.
The NCSC also found that the uplift is uneven:
- Capable state actors and organised crime groups are best placed to use AI in advanced operations, because that needs quality training data, expertise and money.
- Less-skilled criminals, hackers-for-hire and hacktivists gain the most in reconnaissance and social engineering. AI lowers the barrier to entry for them, which the NCSC expects to contribute to the global ransomware threat.
The assessment is from early 2024 and looks two years ahead, so its forecasts cover the period we are now in. Its central message has aged well: AI makes the attacks businesses already face more frequent and more convincing.
The change you will notice first: better phishing
For years, staff training has relied on spotting the clumsy email: odd grammar, wrong spelling, a strange tone. The NCSC notes that generative AI can already produce convincing messages and lure documents without the translation, spelling and grammatical mistakes that often reveal phishing.
That shifts the defence away from "spot the mistake" towards controls that work even when someone is fooled:
- Multi-factor authentication (MFA) means a stolen password alone is not enough.
- Verification habits — calling back on a known number before changing bank details or approving a payment — beat any visual check.
- Reporting culture — staff who report a suspicious message quickly limit the damage when someone else clicks.
Five steps for small businesses and MSPs
- Turn on MFA everywhere it is offered, starting with email, Microsoft 365 and every admin account. In the UK, Cyber Essentials v3.3 now expects MFA on every cloud service that supports it.
- Patch quickly. AI helps attackers analyse and exploit weaknesses faster, so the gap between a fix being released and being installed matters more. High and critical updates within 14 days is the Cyber Essentials standard and a sensible baseline.
- Keep offline or immutable backups and test restoring them. If ransomware gets through, a tested backup is the difference between a bad day and a lost business.
- Set clear rules for AI tools at work. Decide which AI services staff may use, and what must never be pasted into them: client data, passwords, contracts, personal data. Prefer business versions with data-protection terms.
- Ask suppliers how they secure their AI. The Guidelines for secure AI system development, published by the NCSC and CISA with international partners, give a practical checklist across design, development, deployment and operation. They are a good basis for questions to any vendor building AI into the tools you use.
What it means for you at home
The same advice protects your personal email, bank and social media accounts. Turn on 2-step verification, starting with your email. And treat urgency as a warning sign: if a message pushes you to pay, click, or share a code right now, stop and check through a route you already know, such as the number on your bank card.
AI on the defender's side
The NCSC is clear that AI also offers substantial opportunities to defenders: it can improve the detection and triage of attacks and help identify malicious emails and phishing campaigns. Many security products already use it. Treat it as a useful layer, not a replacement for the basics above — and be wary of any product that claims AI makes you "secure" on its own.
The bottom line
AI raises the pace and polish of attacks, but it does not change what works. Businesses that enforce MFA, patch on time, keep tested backups and give staff clear rules are in a far stronger position than those chasing the latest AI headline.
This article summarises and comments on the sources listed below. It is general information, not legal or professional advice.