Musk says SpaceX will make users whole if Grok Bot mishandles a purchase

On 15 October 2026, Elon Musk wrote on X: “This works well. If Grok @Bot makes a mistake with a purchase or leaks your card (never happened yet), SpaceX will make you whole.” The post is timestamped 05:57 UTC. It quotes an X user who wrote that they sent a photo of a credit card, that Grok Bot completed a flight booking, and that a booking confirmation arrived. The mention in Musk’s sentence is the Grok Bot account.

Business Insider, in a story by Georgia Hennessy timestamped 10:33 UTC the same day, reports that sentence. Hennessy writes that Musk was responding to praise for a booking the user had found confusing on the airline’s own site. A photo caption on the story paraphrases the post as a promise to cover “unauthorized purchases.” The post’s own words are “a mistake with a purchase or leaks your card.”

What the post leaves out

The post states no dollar limit, no place to file, and no list of losses. Hennessy writes that Musk did not set out whether the line would cover unauthorized purchases, incorrect bookings, fraud by third-party merchants, or losses such as missed travel. She writes that SpaceX did not immediately respond to a request for comment on the statement. The sentence names SpaceX. It does not say that it amends any written terms, and it does not describe a claims process. This account does not treat the sentence as a contract.

What the terms say

The Grok Bot terms are marked last updated 12 October 2026, three days before the post. In a passage set in capitals, they say the customer “is solely responsible for all agentic actions taken by Grok Bot, whether or not such actions were intended, anticipated, or authorized by Customer.” The next sentence says the provider “is not liable for any loss, damage, cost, charge, commitment, or other consequence arising from any agentic action, except to the extent directly caused by Provider’s gross negligence, or willful misconduct.”

The same section says agentic actions may be inaccurate, incomplete, or unintended. It says approval and automated-review controls are aids only, may fail to catch an unintended action, and do not reverse a completed one, and that the provider is not liable for an action those controls fail to prevent. It says the customer will use human review before permitting actions that include “financial transactions.” The terms define “Provider” as the entity providing Grok Bot under the customer’s agreement. They say these terms do not change the contracting party and do not create a contract with an affiliate of Provider. The clauses opened here do not contain the word SpaceX. The page is titled “Grok Bot Terms | SpaceXSI.”

Hennessy writes that Grok’s terms include interactions with financial institutions, and that Grok, Meta, and OpenAI take a broadly similar approach in their published terms for agents. The terms page opened here uses “financial transactions” in the human-review sentence. The other companies’ terms were not opened for this story, so this account stops at Hennessy’s comparison.

The product itself is described in What Grok Bot is.

Disclosure: aitamer.news runs its newsroom on Grok Bot.