Coinbase says its artificial intelligence system prevented $4.7 million in cryptocurrency transfers to scammers behind a fake Tesla giveaway featuring an AI-generated version of Elon Musk. According to the company’s blog post, The exchange reported that automated defenses stopped approximately 91% of attempted payments during the incident, without requiring human intervention.
The fraud emerged during Tesla's flying car demonstration in October, when scammers launched a deceptive YouTube livestream designed to resemble the company's official broadcast. The stream used a deepfake of Musk to promote a cryptocurrency promotion promising to double deposits made by viewers.
Participants were instructed to send digital assets to wallet addresses displayed on screen or scan a QR code, with the promise that their deposits would be returned at twice the original amount.
Coinbase reported that the livestream attracted more than 500,000 viewers. Within roughly two hours, hundreds of its customers attempted to send millions of dollars to wallets controlled by the fraudsters.
The exchange stressed that the transactions were initiated and authorized by customers themselves. The incident did not involve stolen passwords, compromised account credentials or a breach of Coinbase's security infrastructure.
How Coinbase AI Detected the Cryptocurrency Giveaway Scam
Coinbase used machine-learning models to identify suspicious payment activity before the transfers were completed. The system assessed transaction patterns, wallet characteristics and other signals associated with fraudulent cryptocurrency schemes.
Among the warning signs were recently created wallet addresses, the impersonation of well-known brands and promises of unusually high returns. When suspicious transfers were detected, a dedicated AI scam-prevention agent examined the circumstances surrounding the payments.
The automated agent also communicated with customers, explaining potential warning signs and evaluating the claims used to promote the giveaway. These interactions helped expose misleading promises and discourage users from sending funds to fraudulent addresses.
Coinbase said the specific scam had not appeared in the system's training data. Instead, its models identified familiar fraud indicators in a new scheme built around a major corporate event.
The warning signals included promises to multiply deposits, limited participation windows and the irreversible nature of cryptocurrency transfers. According to the exchange, its automated defenses operated without employees having to review individual transactions or authorize interventions.
The system ultimately blocked approximately 91% of the attempted payments associated with the fraudulent livestream, preventing $4.7 million from reaching the scammers.
Coinbase Plans to Expand AI-Based Scam Prevention
Following the incident, Coinbase said it plans to continue investing in artificial intelligence research and strengthen its ability to detect emerging fraud methods, including scams that use AI-generated content.
The case highlights how fraudsters can use recognizable public figures and high-profile corporate events to persuade people to authorize cryptocurrency transfers. Because customers voluntarily approve these payments, the activity can be difficult to distinguish from legitimate transactions through conventional account security measures alone.
Coinbase's response focused on identifying suspicious payment behavior and warning customers before funds left their accounts. The exchange said the intervention prevented $4.7 million in transfers linked to the fake Tesla giveaway.
Writer: Marcus Renfield Crypto Market Analyst & Onchain Writer
Marcus Renfield covers cryptocurrency markets with a focus on onchain data, Bitcoin price action, and emerging market narratives. His writing examines how capital flows, network activity, and broader market structure influence short- and medium-term trends.
He aims to provide clear, data-informed analysis for readers seeking a deeper understanding of crypto market dynamics.



