Seeing is No Longer Believing with AI – Cyber Security Advice from the Met Police
“ I just got off a video call with the boss, he needs that invoice paid straight away.”
AI is no longer just a buzzword or something from a sci-fi film, it can be an everyday tool for modern business. While it can help us work faster and streamline tedious tasks, it’ s also given cyber criminals a whole new set of tools. You’ re no longer just looking out for poorly written scam emails with obvious typos; you’ re up against clever, AI-driven tricks designed to catch you off guard and slip past your defences. Its important to remember that the foundations of the same scams that we have been fighting for years are still being used but with the arrival of AI, it has allowed criminals to execute these at a much more polished and efficient rate.
Supercharged Deception In the past, Business Email Compromise( BEC) relied on someone hacking an account or spoofing an address. You could usually spot a fake because the spelling was terrible, the formatting looked weird, or it just didn’ t sound like the person who supposedly sent it. AI has wiped out those giveaway clues. Scammers are now using AI language models to write nearperfect emails that match the tone, vocabulary, and style of your manager, supplier, or colleague perfectly.
And it doesn’ t stop at text. Vishing( voice phishing) has had a serious upgrade thanks to voice cloning. A fraudster only needs a few seconds of someone’ s voice taken from a social media clip, a podcast, or a company video to make a synthetic clone. Imagine your accounts team getting a call from someone who sounds identical to your managing director, sounding flustered and asking for an urgent transfer to a new supplier. Would they second-guess it?
The Illusion of Video The newest and perhaps more unsettling trick in the hacker’ s arsenal are real-time deepfakes on video calls. Criminals can now swap faces and voices during live online meetings. There have already been high-profile cases where employees joined a video call, saw and heard who they thought was their finance team, and were convinced to transfer massive sums of money. Everyone else on the screen was a deepfake. When seeing and hearing someone live on screen is no longer proof of who they are, our day-to-day checks have to change.
Never Put AI in Full Control
As more software offers to automate our workflows, there’ s a real danger of getting complacent and relying on it too much. A good rule to remember: never let AI control entire systems on its own, and never blindly trust what it outputs.
You always need a‘ human in the loop’ to check and verify. If an automated system says a payment or a new user account is safe, a real person should still double-check it. If an AI tool suggests changing your network settings or granting new access rights, a human needs to hit approve. Giving total control to automated AI systems is just handing criminals an easy way in.
How to stay ahead
To protect your business from these AI-enhanced scams, solid habits are everything:
• Double-check and verify: If a request comes in for an urgent payment or sensitive data, verify it through a completely different route. Email came through? Call them back on a number you already know and trust.
• Set up safe words: For major financial moves, use internal safe words or phrases that staff must state over the phone or on video before anything gets approved.
• Stick to your processes: Never bypass double-sign-offs or payment limits just because the request seems to come from the top or feels urgent.
For more information on MPS Cyber Protect training go to www. met. police. uk / cyberprotect or email cyberprotect @ met. police. uk