Elon Musk is once again shocking the public, this time with a promise of 100% compensation for losses. According to him, if the AI agent Grok, while managing a client's bank account, makes a financial mistake, the company xAI will fully reimburse the losses. It sounds like a revolution in trust in artificial intelligence, but upon closer inspection, the situation turns out to be far less clear-cut.
Experiment with the Grok Bot Bank Account
An investor under the handle Teslaconomics decided to test the bold claim in practice. In his X account, he asked whether anyone had connected Grok to a real bank account. According to him, such an agent can already track expenses, pay bills, and identify suspicious charges. However, his interlocutor expressed justified concerns: giving AI access to money is a direct path to disaster.
Musk, however, has no doubts about his product. He publicly promised to reimburse losses if the bot makes a mistake. The beta version of the product was released on August 11. At xAI, they explained that each agent operates around the clock on its own cloud server, mimicking human actions, even while the owner sleeps.
This experiment is part of Musk's global financial strategy. Recall that in June, the X Money service with direct transfers between users was already launched. Obviously, AI agents are the next step in this ecosystem.
Grok Bot Liability Limit — $100
Musk's promise looks like insurance, but the company's legal documents say otherwise. In the xAI user agreement, it is written in black and white: the results and actions of agents are provided "as is." The maximum claim amount is limited to the paid fees or $100 — whichever is greater.
Given that access to the bot costs $30 per month under the SuperGrok plan (or $360 per year), this "cap" is just a drop in the ocean compared to potential losses in 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 personal discretion.
The risks are also compounded by US banking rules. Regulation E — a federal standard for electronic transfers — protects customers from unauthorized debits. But if the owner themselves gave the bot access to the account, such a debit is no longer considered unauthorized. Fraud protection mechanisms simply will not work in this case.
Skeptics also point to recent incidents. In May, a malicious NFT "hid" instructions that forced the AI to transfer money — this is called "prompt injection." As a result of an attack on the Bankr wallet linked to Grok, about $150,000 was withdrawn. Later, approximately 80% of the stolen funds were recovered. A few weeks later, 14 more wallets on the same platform were affected, and Bankr promised to fully reimburse the losses. However, there is not a single proven case where Grok made a mistake with a real bank account yet.
If a traditional bank account, rather than a crypto wallet, comes under attack, the damage could be far more serious. The experiment with the Grok bot is partly also advertising: at xAI, they want to integrate the product into everyday financial management. The first real mistake will show what Elon Musk's words are worth. The company will be able to respond in two ways: either silently return all the money, or publicly compensate that very $100 cap.
My analysis: This is a classic example of a marketing promise that is not backed by legal force. Until the guarantee is fixed in the contract, trusting an AI agent with significant sums is an extremely risky move. The market will remember this precedent, and if Musk does not keep his word, the reputational damage to xAI will cost far more than any compensation.