Bankr, a cryptocurrency trading assistant that leverages artificial intelligence to automate transactions, has temporarily disabled its platform after a security breach compromised at least 14 user wallets. The attack, which appears to have been orchestrated through a sophisticated social engineering scheme, resulted in losses reported as high as $150,000 per wallet. In a statement released on X (formerly Twitter), Bankr confirmed that it had identified the attacker and temporarily locked down all transaction activity—including swaps, transfers, and deployments—while the investigation continues.
The company assured users that it would reimburse any and all lost funds. “We've identified an attacker was able to access 14 Bankr wallets. We've temporarily locked things down while we work through the details. We will be reimbursing any and all lost funds,” Bankr wrote. The platform, which allows users to interact with AI in plain language to trade, transfer, and launch tokens, automatically creates a crypto wallet for every X handle that interacts with its bot. This unique feature has raised questions about the security of automated wallet generation and the potential attack surface for malicious actors.
Social Engineering Attack Likely Responsible
Yu Xian, founder of blockchain security firm SlowMist, analyzed the incident and concluded that it was likely a social engineering exploit targeting the trust layer between automated agents. Specifically, the attack involved a manipulated interaction between Grok—a popular AI chatbot—and Bankrbot that allowed unauthorized transaction signing. “It was a social engineering exploit targeting the trust layer between automated agents—specifically an interaction between grok and Bankrbot that allowed unauthorized transaction signing,” Xian explained. He also noted that three identified attacker addresses collectively hold approximately $440,000 in crypto.
This is not the first time such a vulnerability has been exploited. Earlier in 2026, a similar attack had reportedly tricked Grok into requesting that Bankr launch a token, after which funds were drained into a wallet controlled by the hacker. The recurrence of this pattern suggests that the underlying security architecture of AI-integrated crypto platforms may have systemic flaws. Gabi Urrutia, field chief information security officer at Halborn, pointed out that the true cause will remain unclear until Bankr releases a post-mortem, but he outlined several possibilities. These include a compromise of the custodial wallet layer, a session-token issue, or a key-management problem. “Bankr provisions wallets through a third-party custody service, and a key-management or session-token issue there would look exactly like this, with multiple unrelated users drained and no individual phishing footprint,” Urrutia said.
David Schwed, former chief information security officer at Robinhood and current COO at SVRN, emphasized that such incidents often arise from “the glue code between trusted infrastructure and untrusted input.” He noted that wallet compromises can stem from various vectors, including cryptographic flaws like the GG20 threshold signature scheme vulnerability that recently hit THORChain, seed phrase exfiltration via malware, or compromise of programmatic signing infrastructure. “It runs the gamut,” Schwed added, highlighting the complexity of securing decentralized platforms.
Immediate Actions and User Recommendations
In response to the breach, Bankr has advised all users to avoid signing any transactions until further notice. The company warned one individual that their seed phrase “is likely in the hands of an attacker.” For those whose wallets have been compromised, Bankr recommends immediately ceasing use of the affected wallet, generating a new seed phrase on a clean device, and creating a fresh wallet. Users are also urged to move any remaining tokens or nonfungible tokens to the new address and to revoke all approvals if remaining assets cannot be moved. “Attackers often use existing approvals to drain funds. Check your devices, scan your computer and phone for malware or suspicious browser extensions. If you used a software wallet, the leak likely came from your device,” Bankr added.
Industry Implications and Broader Context
The breach comes amid a wave of high-profile crypto heists. In the first quarter of 2026, more than $168.6 million was stolen in crypto-related attacks. April saw two of the largest exploits of the year so far: the $280 million Drift Protocol exploit and the $292 million Kelp exploit. More recently, Verus Protocol's Ethereum bridge was compromised on May 19. The Bankr incident underscores the growing risk faced by artificial intelligence-driven financial tools, which combine the complexities of blockchain security with the unpredictability of AI prompt injection attacks. Security experts are increasingly concerned about the “agent-to-agent” trust models where one AI (like Grok) can influence the actions of another (like Bankrbot), potentially leading to unauthorized fund transfers.
Tech entrepreneur Austen Allred, whose Kelly Claude AI assistant project had a Bankr wallet compromised, confirmed that the hacker stole Ether but left the project’s memecoin stash untouched. “There’s no evidence anyone other than myself ever logged into the Bankr account; they must have accessed the keys some other way,” Allred noted. This statement adds weight to the theory that the attack exploited a vulnerability at the infrastructure level rather than individual user negligence. As the crypto industry continues to integrate AI into trading and wallet management, incidents like this serve as a stark reminder of the security challenges that lie ahead. Developers must prioritize robust auditing, secure key management, and real-time threat detection to protect users from evolving attack vectors.
Source: Cointelegraph News