Haotian AI turns any scammer into anyone, instantly, on your screen 🎭

Cliquez ici pour lire en français

It’s 2pm. Your phone buzzes. A video call — a familiar face, a voice you recognize. You pick up. You talk. You trust. Except the person on the other end isn’t who you think they are.

That’s the reality documented in an investigation published on May 7, 2026, by 404 Media. Reporter Joseph Cox obtained, installed, and tested Haotian AI in live conditions — a real-time deepfake software explicitly marketed to scammers. What he found should concern all of us.

When your face becomes a weapon 🎭

fundamentally different: it lets a scammer replace their face with anyone else’s in real time, during a live call on WhatsApp, Zoom, Teams, TikTok, Instagram, or YouTube. The CyberSignal

When Cox tested the software, the operator’s face morphed into his own — his stubble, his grin, the bags under his eyes. Natural gestures, a hand on the cheek, moving eyes, were all there. The illusion never broke once. 404 Media

The conclusion is hard to escape: a video call is no longer proof of identity.

An industrialized tool, not a prototype 🔧

Haotian AI is likely built on open-source face-swap tools that have been publicly available for years. The commercial value its sellers provide is technical support — installation, configuration, an interface for adjusting nose size or eye shape, and operational guidance for targeting specific platforms. The technology isn’t novel. What’s new is the packaging, the accessibility, and the support ecosystem around it.

Blockchain analytics firm Chainalysis found that wallets linked to Haotian AI have processed more than $4 million in total inflows since October 2023. Its Telegram channel, launched at the same time, has surpassed 20,000 subscribers and serves as a hub for updates and technical support. This is a business — structured, profitable, and growing.

The AI sees nothing 👁️

What makes this threat worse is the failure of existing detection systems. Xception, one of the most studied academic deepfake detectors, achieves 89.1% accuracy on standard control samples — but misclassified nearly 100% of Haotian-generated outputs as « authentic. » Detection tools were built to catch the previous generation of deepfakes. Haotian bypasses them almost entirely.

What this means for Africa 🌍

It would be easy to frame Haotian AI as a distant problem. That would be a mistake.

WhatsApp is the dominant communication platform across sub-Saharan Africa — used daily for financial transactions, professional exchanges, and family contact. Haotian AI runs natively on it. Documented use cases include romance scams, virtual kidnappings, tax fraud, and business email compromise — schemes already prevalent across the continent, now paired with a tool that makes them visually irrefutable.

Trust in the deepfake era 🔐

Haotian AI signals a shift. Analysts expect rapid proliferation of competing real-time deepfake tools throughout 2026 and 2027. The question is no longer whether these tools will spread, but how fast. The CyberSignal

Three habits worth adopting now: never validate a sensitive transaction based solely on a video call; establish verification codes with close contacts for emergency situations; and flag unusual behavior even when the face on screen looks familiar.

Real-time deepfakes run on a laptop, travel through WhatsApp, and are already active. The best defense is knowing that.

💬 How do you protect yourself from online scams? Have you ever faced an identity impersonation attempt? Share your experience in the comments — the TechGriot community is reading.