2 min read
Add as a preferred source on Google

Coinbase AI Blocked $4.7 Million in Musk Deepfake Scam Transfers

More than 500,000 people watched the fake Tesla flying-car demonstration, while hundreds of customers tried to send crypto to addresses shown on screen.

A thumb pauses above a transfer control on a smartphone / TokenPost.ai
A thumb pauses above a transfer control on a smartphone / TokenPost.ai

Coinbase’s fraud defenses blocked about $4.7 million in attempted transfers during a fake Elon Musk crypto giveaway livestream, stopping roughly 91% of customer funds over two hours.

The broadcast showed an AI-generated version of Musk during what appeared to be a Tesla flying-car demonstration. Wallet addresses and a QR code appeared on screen alongside a promise to double any crypto sent. More than 500,000 people watched, and hundreds of customers tried to transfer funds.

No Coinbase accounts were breached, and no passwords or login credentials were stolen. Customers signed in and approved the transfers themselves; many were longtime users. Because the customers authorized the payments, the activity differed from transactions fraud systems typically flag as unauthorized.

Coinbase’s machine-learning models scored transfers in real time, looking for patterns such as giveaways, brand impersonation and newly created wallet addresses. When a transfer was flagged, an AI agent explained the suspected scam and asked the customer questions. The system also scored the agent’s replies to help decide when to add another step before a transfer could proceed.

The agent had not previously encountered a deepfake giveaway tied to a product-launch livestream. It recognized several warning patterns: a promise to multiply funds, pressure to act before the event ended and transfers that could not be reversed. Those signals helped it handle the flagged transfers from the first through the last.

Coinbase is investing in AI for its risk operations. Its plans include using AI research to update detection models, deploying agents to review high-risk transfers across scam types, and labeling new cases so models can learn from attacks often within hours instead of weeks.

Simon Yoon

Reporter

Simon Yoon reports on blockchain technology for TokenPost. Send corrections or tips to info@tokenpost.com.

Loading…