Elon Musk says SpaceX will pay back anyone whose Grok AI agent makes a rogue purchase

• Elon Musk says SpaceX "will make you whole" if your Grok AI agent makes unauthorized purchases.
• Musk's comment, though light on detail, is a departure from the standard approach to agent mistakes.
• Grok, Meta, and OpenAI's terms of service generally place responsibility for agent actions on users.

Elon Musk said that SpaceX will pay users back if Grok’s AI agent makes a mistake while completing a purchase or leaks their card information.

In a Thursday X post, Musk said the company "will make you whole" if Grok makes a financial mistake. He made the comment in response to an X user praising Grok for booking them a flight after being sent a photograph of their credit card.

The user described the airline’s website as confusing and said Grok completed the transaction and delivered a booking confirmation.

"This works well," Musk wrote. "If Grok Bot makes a mistake with a purchase or leaks your card (never happened yet), SpaceX will make you whole."

Grok, Meta, and OpenAI all take a broadly similar approach in their published terms for agents: users are responsible for the actions and transactions their AI agents carry out on their behalf.

Meta says Muse requires approval before purchases, while Grok’s terms include interactions with financial institutions, but none offers the kind of explicit reimbursement pledge Musk described.

Musk did not set out the scope of the apparent guarantee, including whether it would cover unauthorized purchases, incorrect bookings, fraud by third-party merchants, or consequential losses such as missed travel.

SpaceX did not immediately respond to a request for comment on Musk's assertion.

The risks are already emerging as agents gain access to payment methods and online accounts.

In a recent example involving Grok, Shane Mac, the CEO of software company XMTP Labs, told Business Insider's Aditi Bharade that a personal-finance agent he created posted his bank balances and spending details in his company’s executive Slack channel after confusing two similarly named chats.

Grok’s team later changed its permission controls, according to Mac.

Similarly, one user of the AI agent Instinct said it ordered 20 pounds of chicken wings during an agent-assisted grocery shop, while a Muse user said the agent accepted a low offer for a keyboard on Facebook Marketplace and shared his home address with the buyer.

He later said he had selected an "Allow Always" setting, though he had believed the agent would still seek approval before accepting offers.

While Musk’s promise is short on detail, it raises a broader question for the companies building agents that can spend money and access sensitive data: Will users continue to be expected to bear the risk, or will the companies behind the tools begin to share it?