CYBERSECURITY
The people who use AI most effectively for harm are not governments or corporations. They’re scammers.
Fraud has always been an early-adopter industry. When telephony became ubiquitous, phone scams followed. When email became universal, phishing followed. When social media matured, social engineering at scale followed. AI is simply the latest technology to be weaponised, and the scam ecosystem has adopted it faster and more creatively than most legitimate organisations.
What We’re Seeing Now
AI voice cloning has moved from experimental to commodity. A three-second audio clip from a social media video is sufficient to clone a convincing voice. Fraudsters are using this to impersonate family members in “grandparent scams,” to fake executive voices in authorisation fraud, and to create synthetic customer service agents that harvest credentials.
Real-time deepfake video for video calls is more expensive but increasingly accessible. We’ve seen documented cases of HR teams being fooled by convincing deepfakes during remote job interviews — resulting in fraudsters being hired, granted system access, and used to exfiltrate data or install malware.
“We’ve seen documented cases of HR teams being fooled by convincing deepfakes during remote job interviews.”
What’s Coming
The next wave will be personalised phishing at scale. AI can now ingest a target’s public digital footprint — social media, professional profiles, company websites, public records — and generate a bespoke phishing communication that references specific details with a specificity that previously required hours of manual research per target. Automated at scale, this changes the economics of spear phishing entirely.
The defences need to evolve in parallel. Multi-factor verification for voice communications, out-of-band confirmation for unusual requests, and systematic scepticism about urgency are the practical countermeasures. The paranoia that used to be excessive is now appropriate.
Tags: Artificial Intelligence • Opinion • Technology & Society • 192.168.1.22/