Digital Asset
Intelligence
Theme

Search 3CBlock

News, exchanges, regulatory intelligence, reviews and Voices.

Coinbase AI Blocks $4.7 Million in Deepfake Tesla Livestream Scam

Add us on Google

Summary: Coinbase’s automated agent intercepted nearly 91% of attempted transfers during a fraudulent YouTube broadcast featuring a deepfake Elon Musk, preventing approximately $4.7 million in losses without human intervention.

Credibility editorial illustration by 3CBlock
Illustration: 3CBlock

Coinbase blocked approximately $4.7 million in cryptocurrency transfers destined for scammers operating a fake Tesla livestream. The fraudulent broadcast, which featured a deepfake representation of Elon Musk, attracted more than 500,000 viewers and promised to double any crypto sent to provided wallet addresses. The exchange’s artificial intelligence agent stopped nearly 91% of the attempted payments with no human review, intercepting the funds before they left customer accounts.

Deepfake Broadcast Mimics Official Event

Scammers established a YouTube livestream during Tesla’s “flying car” demo that closely resembled the official broadcast. The fake stream displayed a deepfake of Musk, along with wallet addresses and a QR code, pledging to double any funds sent by viewers. More than half a million people tuned in to the deceptive presentation. Hundreds of Coinbase customers, including many long-time users, attempted to send funds within a two-hour window, believing the giveaway was legitimate.

Coinbase clarified that no accounts were hacked and no passwords or login credentials were stolen during the incident. Customers approved the transfers themselves because they believed the giveaway was real, a scenario that slipped past traditional fraud tools built to catch unauthorized activity rather than authorized but deceived transactions.

AI Agent Intervenes Without Human Input

The exchange’s existing AI-powered scam-prevention system utilizes a three-layer process to assess suspicious transfers. Machine learning models score transactions in real time, looking for giveaway patterns, brand impersonation, and transfers to new wallet addresses. Once a send is flagged, the agent initiates a short conversation with the customer, identifies the scam pattern, explains how giveaway schemes work, and asks questions that expose the inconsistencies in the fraudulent narrative. Each reply carries a confidence score.

According to Coinbase, the agent had never previously encountered a deepfake giveaway tied to a live product launch. Despite this novelty, the system caught the scheme and operated from the first flagged send to the last with no humans in the loop. In its analysis of the event, the exchange stated, “Nobody legitimate will double your crypto. An offer that requires you to send funds first to receive more back is a scam, no matter who appears to be making it.” The company urged users not to send crypto to any address or QR code shown in a video, livestream, or message.

Rising Threat of AI-Generated Fraud

Coinbase noted that staging a scam of this magnitude once required a crew, a studio, and weeks of work. Now, a livestream slot and an AI model are sufficient, allowing operators to relaunch under a new name as soon as one stream is taken down. Musk deepfakes are not new; a 2024 New York Times report found thousands of AI-generated videos featuring phony versions of Musk, Warren Buffett, and Jeff Bezos deceiving would-be investors. Deloitte estimates that AI-powered deepfakes could drive billions of dollars in fraud losses each year.

In response to these evolving threats, Coinbase is investing heavily in AI across its risk operations. This includes agents reviewing risky sends across every scam type and AI labeling that trains its models on new cases within hours. The successful interception of the $4.7 million in potential losses highlights the growing role of automated systems in protecting users from sophisticated social engineering attacks that bypass traditional security measures.