Coinbase AI blocks $4.7 million in deepfake Elon Musk giveaway scam
Coinbase’s AI‑driven security system stopped roughly $4.7 million from being sent to scammers who used a deepfake Elon Musk livestream.
Key takeaways
- Coinbase’s AI defenses intercepted about 91 % of $4.7 million in attempted transfers during the scam.
- The fake livestream attracted more than 500,000 viewers before the QR‑code donation request appeared.
- TRM Labs reported deep‑fake‑related crypto losses in 2026 were up 263 % versus 2025.
Coinbase announced that its artificial‑intelligence‑powered fraud detection system blocked roughly $4.7 million from being transferred to scammers who used a hyper‑realistic deepfake of Elon Musk during a hijacked YouTube livestream of Tesla’s “flying car” demonstration. The AI intervened within two hours of the scam’s launch, preventing the majority of the attempted transactions without human involvement.
How the scam unfolded
The attackers took over the livestream that was promoting Tesla’s upcoming flying‑car prototype and inserted a fabricated video of Musk promising to double any crypto sent to a displayed QR code. More than half a million viewers tuned in, and within minutes hundreds of Coinbase users attempted to send funds to the scammers’ wallets.
AI‑driven defense in action
According to Coinbase, its automated security stack identified the suspicious QR‑code transaction pattern and halted about 91 % of the attempted transfers. No manual review was required, marking a shift from earlier incidents where the exchange had to blacklist addresses after the fact, such as the 2020 double‑your‑money scam involving a hacked Musk‑related tweet.
Context of rising deepfake scams
TRM Labs data cited by Cryptopolitan shows that reported losses from deepfake‑driven crypto scams in 2026 have already exceeded 2025 figures by 263 %, underscoring the rapid growth of this threat vector. While overall crypto thefts this year total $8.78 billion across 343 incidents, scams now outpace traditional hacks in speed and scale, according to DeFiLlama.
What to watch next
Industry observers will monitor whether other exchanges adopt similar AI defenses and how regulators respond to the increasing sophistication of deepfake‑based fraud. The next major test for Coinbase’s system could come as deepfake technology becomes cheaper and more accessible, potentially prompting broader industry standards for automated scam prevention.
FAQ
What was the amount blocked by Coinbase’s AI?
The system intercepted roughly $4.7 million in attempted transfers.
How many viewers watched the fake livestream?
More than 500,000 people tuned in to the hijacked stream.


Discussion 0