Hyperliquid Strategies' HYPE Profit Cushion Shrinks to Pennies
Hyperliquid Strategies, the largest Hyperliquid treasury, now holds HYPE at a profit of only a few cents per token after a 28% decline in price.
Coinbase's AI tool blocked $4.7 million in customer funds from a deepfake Elon Musk crypto giveaway scam.
Coinbase (COIN) announced that its AI-powered scam prevention tool prevented approximately $4.7 million in customer funds from being sent to fraudsters who used a deepfake video of Elon Musk to promote a fake Tesla cryptocurrency giveaway. The system stopped about 91% of the money customers attempted to transfer. The exchange detailed the event in a Tuesday blog post as part of a recurring consumer protection series, noting the system operated “with no humans in the loop.”
These figures indicate customers attempted to send roughly $5.2 million, with around $460,000 still leaving their accounts. Coinbase did not disclose how much reached the scammers.
The scam occurred during Tesla’s “flying car” demonstration. Fraudsters created a YouTube livestream resembling the actual broadcast, featuring a deepfake Musk, on-screen wallet addresses, and a QR code. The stream promised to double any cryptocurrency sent to it, attracting over half a million viewers, according to Coinbase.
Within roughly two hours, hundreds of Coinbase customers attempted to send millions of dollars to the displayed addresses, many of whom were long-time users. Coinbase did not specify when the fake stream took place or which cryptocurrencies were involved.
The company emphasized that its platform was not compromised. “Nobody was hacked,” Coinbase wrote. “No Coinbase account was broken into, no password stolen, no credential phished.” This makes giveaway fraud difficult to prevent because most defenses target unauthorized activity, whereas in this case customers logged in and confirmed transactions themselves.
“Our systems can see risk signals in real time. They can’t see intent,” the post stated.
Coinbase described a three-part defense system. Machine learning models first score outgoing transactions in real time, looking for giveaway patterns, brand impersonation, and newly created wallet addresses. The company wrote that the aim is “to catch the danger, not the customer.” When a transaction is flagged, a scam-specific agent takes over, gathering context, identifying the scam type, explaining how giveaways work, and asking questions designed to expose the fraud.
Each reply is scored for confidence before the customer sees it, ensuring the system knows when more friction is needed. Coinbase noted the agent had never been specifically trained on this type of con, as deepfake giveaways tied to a live product launch were not in its training data.
“It recognized the shape of the scheme anyway,” the company wrote, citing the promise to multiply funds, the deadline of a live event, and an irreversible transfer.
Coinbase stated that deepfakes have altered the scam’s economics. Staging such an operation once required “a crew, a studio, and weeks,” it wrote. “Now it takes a livestream slot and a model.” The company said it is expanding AI review of risky transactions to all scam types, not only high-profile ones, and is also using AI to label new scam cases so models learn from attacks within hours instead of weeks.
Social engineering has been a persistent issue for Coinbase customers. In September, a Brooklyn man received a 12-year prison sentence for a $16 million Coinbase phishing scam. This week, Trezor customers reported scam calls after a breach at the hardware wallet maker’s shipping partner exposed 27,000 orders.
The blog post coincided with Coinbase launching a ChatGPT plugin for trading bitcoin, stocks, and derivatives, which requires user approval for every order. The post concluded with a list of actions Coinbase will never take, including promising to double, triple, or multiply cryptocurrency sent by customers.
“Nobody legitimate will double your crypto,” the company wrote.
Share to
Disclaimer: this article comes from third-party media and is provided for reference only. It does not constitute investment advice. Crypto and other financial products carry significant price volatility risk, so please make your own decisions carefully.
Hyperliquid Strategies, the largest Hyperliquid treasury, now holds HYPE at a profit of only a few cents per token after a 28% decline in price.
Bitcoin fell below $57,000, losing 30% in a month. Analysts predict a bottom between $50,000 and $54,000.
Crypto liquidations reached $1.76 billion in 24 hours, involving Bitcoin, Nvidia, and SanDisk bets, as Nvidia fell to fourth and Bitcoin dropped below $60,000.
Bitcoin fell to about $58,300, down 6.2%, as a chip stock rout sent the Nasdaq sliding 5%.