Community Trust ScoreVerified
Coinbase reported that its automated fraud-detection system intercepted $4.7 million in attempted cryptocurrency transfers destined for a fraudulent scheme. The scam involved a deepfake video of Elon Musk promoting a fake Tesla crypto giveaway. The company stated that its AI-powered defenses stopped these funds from reaching the scammers, preventing approximately 91% of the linked transfers without any human intervention.
Fake Tesla Livestream
The incident unfolded during Tesla’s “flying car” demonstration last month. Scammers created a YouTube livestream designed to mimic Tesla’s official broadcast. The stream featured a deepfake of Elon Musk and displayed on-screen wallet addresses alongside a QR code. Viewers were told that any cryptocurrency sent to these addresses would be doubled. More than 500,000 people watched the fraudulent stream. Within roughly two hours of the broadcast, hundreds of Coinbase customers attempted to send millions of dollars to the scam addresses.
AI Detection Without Hacks
Coinbase emphasized that the success of the block was not due to a security breach. “The part that makes this hard: nobody was hacked,” the company stated. “No Coinbase account was broken into, no password stolen, no credential phished. Hundreds of customers, many of them long-time users, decided on their own to send money to a stranger.” The exchange noted the inherent difficulty in stopping transactions that customers are tricked into authorizing voluntarily.
Coinbase’s defense mechanism relies on machine-learning models that score risky sends in real time. These models flag transactions that match patterns associated with giveaways, brand impersonation, and newly created wallet addresses. When a transaction is flagged, a dedicated scam-prevention agent is dispatched to examine the transaction context. This agent explains how such scams are structured and asks questions to deconstruct the scammer’s narrative. In this instance, the AI agent had not encountered this specific type of scam during its training. “What’s notable is that the agent had never seen this scam. Deepfake giveaways tied to a live product launch weren’t in its training,” Coinbase said.
Despite the lack of specific training data, the system identified the fraud. “It recognized the shape of the scheme anyway, a promise to multiply funds, the urgency of an event that would end in hours, a transfer that could not be reversed, and it ran from the first flagged send to the last, with no humans in the loop,” the company explained. The AI system operated autonomously throughout the incident, intercepting the bulk of the attempted transfers. Coinbase stated it will continue to invest heavily in AI defense, aiming to create models that keep pace with new scam types through AI-driven research.