Worried About AI-Powered Cyber Threats? Get the Basics Right First

computer screen and glasses

AI is changing how attacks are carried out. It’s also changed remarkably little about how to defend against them.

The most effective response to AI-powered threats isn’t a new, exotic piece of AI-detection software — it’s making sure the fundamentals that have always mattered are actually in place. Most successful attacks, AI-assisted or not, still succeed because of the same gaps: no MFA, unpatched software, no proper monitoring, and people who haven’t been trained to spot a convincing fake.

How AI is actually changing the threat landscape

Two changes are genuinely worth knowing about.

AI-written phishing emails are harder to spot. The old advice — look for bad spelling and grammar — is far less reliable than it used to be. AI tools can now write a fluent, well-punctuated, convincingly-worded phishing email in seconds. The old tells are disappearing.

AI voice cloning makes vishing more convincing. Voice phishing attacks increasingly use AI-generated voice cloning to impersonate a real person — a senior colleague, a supplier, a bank representative — making a fraudulent phone call sound exactly like someone your team would normally trust. We cover this in more detail on our mobile phishing page.

Neither of these requires a new category of defence. They require the same foundational controls, done properly, and a team that knows what to watch for regardless of how convincing the message sounds.

The foundation that actually stops most attacks

Multi-factor authentication, enforced everywhere. If a password is compromised — whether through an AI-written phishing email or any other method — MFA stops the attacker getting any further. This remains the single most effective control against account compromise. See our Security page.

Patching and updates, applied automatically. Outdated software is still the most common entry point for attackers, AI-assisted or otherwise. Automatic, tested patching closes known vulnerabilities before they can be exploited.

Antivirus and advanced threat detection. Signature-based antivirus catches known threats. Behaviour-based detection (EDR) catches the unknown ones, including new attack patterns that didn’t exist when your antivirus definitions were last updated. See our Antivirus & Security page.

Web-level threat blocking. Malicious links, however convincingly they were written, still need to connect to a malicious destination. DNS filtering blocks that connection before it loads. See our DNS Security page.

Security awareness training. Technology stops most attacks. A team that knows to verify a request through a separate channel before acting on it — even one that sounds completely convincing — closes the gap that technology alone can’t. See our Cybersecurity Awareness Training page.

Cyber Essentials. The five technical controls behind Cyber Essentials certification — firewalls, secure configuration, access control, malware protection, and patch management — are precisely the foundation that makes a business resilient against new and evolving threats, not just old ones. See our Cyber Essentials page.

The honest takeaway

If your business already has these fundamentals properly in place, you are well positioned against AI-driven threats — not because you’ve bought a specific “AI security” product, but because the underlying defences work regardless of how an attack was generated.

If you’re not certain whether these fundamentals are properly in place, that’s exactly what a free site survey is for.

One monthly fee. One number to call.

The day-to-day risk of keeping up with how attacks are evolving becomes our job, not yours.

Book your free site survey   or call +44 (0) 207 403 4031

FAQ

Common questions

Is AI making cyber attacks more dangerous?

AI is changing how some attacks are carried out, particularly phishing emails (which can now be written more convincingly, without the spelling and grammar errors that used to be a giveaway) and voice phishing, where AI voice cloning can make a fraudulent phone call sound exactly like someone genuine. However, the underlying defences that stop these attacks haven’t changed: multi-factor authentication, patching, antivirus and threat detection, and a team trained to verify suspicious requests still work regardless of how the attack was generated.

Do we need special AI-detection software to protect against AI-powered threats?

Not as a starting point. Most successful attacks, whether AI-assisted or not, succeed because of the same underlying gaps: no MFA, unpatched software, no monitoring, or staff who haven’t been trained to spot a suspicious request. Getting these fundamentals properly in place is significantly more effective than adding a specific “AI security” product on top of a weak foundation.

How can our team spot an AI-written phishing email if it doesn’t have the usual spelling mistakes?

The old advice of looking for poor grammar is far less reliable than it used to be. The more reliable approach is to verify unusual or urgent requests through a separate channel — calling a known phone number rather than replying to the email — rather than relying on spotting obvious errors. Security awareness training helps build this habit across your team.

What is AI voice cloning and how does it relate to vishing?

AI voice cloning technology can now convincingly replicate a specific person’s voice from a small amount of sample audio. Attackers are increasingly using this in voice phishing (vishing) attacks, making a fraudulent call sound exactly like a senior colleague, a supplier, or another trusted contact. The defence is the same as for any vishing attempt: verify any unusual or urgent request through a separate, known communication channel before acting on it.

Does Cyber Essentials certification help protect against AI-powered threats?

Yes. The five technical controls assessed under Cyber Essentials — firewalls, secure configuration, access control, malware protection, and patch management — form the foundation that makes a business resilient against evolving threats generally, including AI-assisted ones. These controls aren’t specific to any particular attack method, which is exactly why they remain effective as attack techniques change.

What’s the single most effective thing our business can do against AI-powered phishing?

Enforce multi-factor authentication across every account. Even if an AI-written phishing email successfully tricks someone into entering their password, MFA stops the attacker getting any further without the second verification step. It remains the most effective single control against the consequences of a successful phishing attempt, regardless of how convincing the original message was.