Dangers of deepfakes

The glitching accent in this TikTok clip is laughable, but do not mistake incompetence for harmlessness. This is just the latest wave of digital 'slop' designed to game algorithms and separate fools from their money.

The real danger isn't a fake podcast host selling skincare; it's the industrialised theft of identity. We are already seeing deepfakes of trusted experts, like the Martin Lewis investment scams, used to defraud the vulnerable out of millions. The barrier to entry for high-level fraud has effectively dropped to zero.

In our AI Audits, we are finding that content verification is becoming a critical business asset. We tell clients: if you cannot prove the human provenance of your content, you have zero credibility.

We are rapidly moving from 'trust but verify' to 'verify or assume it’s a lie.' Until platforms are held liable for hosting this fraud, the technology will continue to be weaponised. It’s not just about scary tech; it’s about the total collapse of digital trust.

Deepfakes are no longer scary, they are now a solid business model and the targets aren't world leaders, they're your friend buying face cream.

That Korean skincare deepfake is just the polished version of what we saw with the "hospital staff" scam peddling fat loss patches a few weeks ago.

When fraud becomes scalable, easy, and platform-approved through TikTok Shop integration et al, you don't have a tech problem. You've got a marketplace that's willingly monetising deception.

These aren't sophisticated attacks requiring nation-state resources anymore. This is consumer-grade fraud-as-a-service. The accent-switching AI voice is a tell now, but give it six months. How long are platforms going to profit from looking the other way.

The AI needed to cause havoc is outstripping the AI moderation bots.