Elon Musk has once again found himself at the center of the crypto community's attention, this time due to his AI creation. In response to user skepticism about the safety of the autonomous agent Grok, which can now manage a bank account, Musk publicly promised to compensate any losses if the bot makes a mistake. It sounds like perfect insurance, but upon closer inspection, xAI's legal documents paint a completely different picture.
One investor in Tesla and xAI, known by the handle Teslaconomics, decided to run an experiment. He asked a simple question: had anyone already connected the Grok bot to their bank account? In a post on X, he noted that the agent can track expenses, pay bills, and even spot suspicious charges. The idea inspired him so much that he wondered whether he even needed a personal banker anymore. However, his colleague expressed concern that an AI with access to finances could cause trouble. Musk was not deterred by such doubts — he is confident in the product and promised to refund money in case of an error.
Beta version and global plans
The beta version of the product was released on August 11. xAI explained that each agent operates around the clock on its own cloud server. It visits websites like an ordinary person and continues working while the owner sleeps. This experiment fits into Musk's global financial strategy: in June, the X Money service with direct transfers between users had already gone live.
Limitation of liability: $100 — and that's it
Musk's promise sounds like a guarantee, but xAI's user agreement explicitly states: the results and actions of agents are provided "as is." The maximum claim amount does not exceed the paid fees or $100 — whichever is greater. At the same time, access to the bot costs $30 per month under the SuperGrok plan, which adds up to $360 per year. That amount is laughable compared to what could be lost from an account.
Musk's response on X does not change these terms. Until xAI formalizes the guarantee in writing, any compensation remains at Musk's discretion. Moreover, U.S. banking regulations only amplify the risks. Regulation E — the federal rule for electronic transfers — protects customers from unauthorized charges. But if the owner themselves gave the bot access to the account, such a charge is no longer considered unauthorized.
Prompt injection and real attacks
Skeptics point to recent incidents. In May, a malicious NFT "hid" instructions that forced an AI to transfer money — this is called "prompt injection." As a result of the attack, about $150,000 was withdrawn from the Bankr wallet linked to Grok. Later, roughly 80% of the stolen funds were recovered. A few weeks later, 14 more wallets on the same platform were affected, and Bankr promised full reimbursement of losses. So far, there is not a single proven case where the Grok bot made an error with a real bank account.
However, if a crypto wallet is not the target, the damage could be far more severe.
The experiment with the Grok bot is partly also advertising: xAI wants to integrate the product into everyday financial management. Notably, the very first real mistake will show what Elon Musk's words are worth. The company can respond in two ways: either silently refund all the money, or publicly compensate that very $100 cap.
My verdict: For now, this is more of a marketing move than a real guarantee. Trusting an AI agent with large sums without clear legal protection is like playing Russian roulette. Until xAI formalizes its obligations in writing, I recommend viewing Grok as a useful tool, but not as a replacement for professional financial management.