An attacker infiltrated the dashboard of the AI-driven crypto bot aixbt, stealing 55.5 Ether ($106,200) by triggering two unauthorized transactions. The exploit, which occurred on March 18, has raised fresh concerns about the security of AI-based trading systems in crypto.

The bot’s maintainer, rxbt, confirmed in an X post that the core system remained intact and the breach did not involve direct AI manipulation. In response, the team migrated servers, swapped security keys, paused dashboard access, and reported the hacker’s addresses to exchanges in an effort to contain the damage.

Following the incident, aixbt’s token (AIXBT) dropped 15.5% to $0.09, according to CoinGecko data. Initially, speculation arose that the AI had been manipulated, especially after Simulacrum AI, another AI-powered platform, posted about sending the stolen funds to X user 0xhungusman, whose account was later suspended.

AI-driven bots like aixbt, ai16z, and Truth Terminal are gaining popularity among crypto traders looking to leverage AI for market insights and automated trading. However, the attack highlights the vulnerabilities of AI-driven finance. Spencer Farrar, a partner at venture capital firm Theory Ventures, acknowledged that while AI in crypto is in its early stages, more practical applications could emerge as the sector matures.

Decentralized AI researcher S4mmy stressed the need for better security testing for AI agents handling crypto assets, predicting that major DeFi protocols will likely integrate or develop AI solutions in the next year.

Currently, AI-related crypto tokens boast a market cap of $4.2 billion, underscoring the growing investment in the sector—despite its security risks.