埃隆·马斯克表示，如果Grok的AI代理在完成购买时出错或泄露用户卡信息，SpaceX将向用户退款。他在周四的X帖子中表示，公司“会让你完整无损”，如果Grok在财务方面出错。他在回复一位X用户时作出了这一评论，该用户称赞Grok在收到其信用卡照片后为其预订了航班。该用户描述航空公司的网站令人困惑，并表示Grok完成了交易并提供了预订确认。
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.
这很有效。如果Grok @Bot在购买时出错或泄露您的卡（目前尚未发生），SpaceX会让你完整无损。马斯克写道：“这很有效。如果Grok Bot在购买时出错或泄露您的卡（目前尚未发生），SpaceX会让你完整无损。”Grok、Meta和OpenAI在其发布的代理条款中都采取了大体相似的做法：用户需对其AI代理代表其进行的行为和交易负责。
This works well. If Grok @Bot makes a mistake with a purchase or leaks your card (never happened yet), SpaceX will make you whole. "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表示Muse在购买前需要批准，而Grok的条款包含与金融机构的互动，但没有一种提供马斯克所描述的那种明确的赔付承诺。马斯克并未说明这一看似保证的范围，包括它是否会覆盖未经授权的购买、错误预订、第三方商家的欺诈行为，或诸如错过旅行之类的连带损失。SpaceX尚未就马斯克的这一说法立即回应置评请求。
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.
随着代理获得对支付方式和在线账户的访问权限，风险已经开始显现。在最近一个涉及Grok的例子中，软件公司XMTP Labs的首席执行官Shane Mac告诉Business Insider的Aditi Bharade，他创建的个人理财代理在将两个名称相似的聊天弄混后，在公司高管的Slack频道中发布了他的银行余额和消费详情。据Mac透露，Grok团队随后修改了其权限控制。
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.
同样，一位使用 AI 代理 Instinct 的用户表示，在代理协助的杂货购物过程中，它订购了 20 磅鸡翅；而一位 Muse 用户表示，该代理在 Facebook Marketplace 上接受了键盘的低价报价，并向买家分享了他的家庭地址。他后来表示，他已选择了“始终允许”设置，尽管他认为代理在接受报价前仍会寻求批准。虽然马斯克的承诺细节不足，但它为正在构建能够花费金钱和访问敏感数据的代理的公司提出了一个更广泛的问题：用户是否仍将被期望承担风险，或者这些工具背后的公司是否将开始分担风险？
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?