Deepfake Fraud Costs Victims $3.7B as Social Media Enables AI Scams

What happened

New research published by cybersecurity firm Surfshark has found that deepfake-enabled fraud has cost victims at least $3.7 billion globally, with social media platforms identified as the primary vector through which these scams originate. The study maps how AI-generated impersonations — spanning fake celebrity endorsements, fabricated video calls, voice cloning and messaging-app deception — move from initial exposure on social platforms to direct financial harm.

Surfshark's findings highlight that social media's combination of high reach, algorithmic amplification and relatively low content-verification standards makes it structurally well-suited to deepfake distribution. Once a fraudulent video or audio clip gains traction, the trust heuristics that make social platforms compelling — familiar faces, peer sharing, apparent authenticity — work against the very users they are meant to engage.

Why it matters

For customer-experience and service-design professionals, this research is a direct challenge to one of the field's foundational assumptions: that familiarity and recognisability build trust. Deepfake fraud exploits precisely those cues — a known face, a familiar voice, a credible brand — to manufacture false confidence and drive harmful action. When the signals organisations invest in to signal legitimacy can be synthetically replicated at scale, the entire architecture of trust-based customer journeys requires re-examination.

From a behavioural-economics perspective, deepfakes are an industrialised form of authority bias and social proof manipulation. Victims are not being careless; they are responding rationally to stimuli that have historically been reliable trust signals. That distinction matters enormously for how brands, platforms and regulators should respond — and for how service designers should think about verification, friction and the moments where slowing a customer down is actually an act of care.

By the numbers

• $3.7 billion — minimum estimated global losses attributed to deepfake fraud, according to Surfshark's research.
• Social media identified as the single largest origination channel for deepfake scam content across all fraud typologies studied.

The Renascence take

The instinct among brands will be to treat deepfake fraud as a cybersecurity problem to be handed off to IT and legal. That framing is too narrow, and it will leave customers exposed at the very touchpoints that CX teams own.

The deeper issue is that decades of CX practice have trained customers to trust polished, familiar, emotionally resonant signals — and deepfake technology now manufactures those signals on demand. The behavioural principle underneath is not gullibility; it is that we have systematically removed friction from trust-formation, and fraudsters are the direct beneficiaries. Customer-obsessed operators should be asking a pointed question right now: at which moments in our customer journey does someone make a high-stakes decision based solely on perceived authenticity? Those moments need deliberate, human-centred verification design — not just a disclaimer, but a genuine pause that gives customers permission to confirm before they commit.