Elon Musk published a post on X boasting about how his Grok Bot AI agent successfully ordered some Lego Star Wars kits. The screenshot he shared with his more than 240 million followers inadvertently exposed his 10-digit Lego order number, pricing, delivery and billing addresses, email address, and the last four digits of his credit card.
The Exploitation of the Lego Order Number
Faster than the Millennium Falcon making the jump to light speed, an X account holder named Enderman posted screenshots showing interactions with Lego’s customer service chatbot on its website. Using the email and order number provided in Musk’s initial post, the user successfully manipulated the system into canceling the order.
Within a couple of hours, Musk replied to test whether the cancellation had actually stuck, stating that his order had not been canceled after all. However, the security lapse opened the door for further interference. Another X user returned to Lego’s AI assistant, named Sophia, and used the exact same credentials to reopen the delivery. This time, the user escalated the purchase volume dramatically.
Scaling the Brick Count to Four Hundred Sets
The second user increased the order count to 400 Lego Star Wars sets and altered the delivery address while keeping Musk’s original payment method active. According to screenshots shared on X, the chatbot responded with compliance: “Perfect! I’ll get your Lego adventure back on track.”
That 400-set escalation remains a fraction of what automated buying systems can achieve. A glance at WIRED’s Spend Elon Musk’s Money tracker demonstrates that blowing $1 trillion of Musk’s cash on plastic bricks theoretically allows for the purchase of one billion $1,000 Death Star sets, each containing 9,023 pieces.
Lego declined to comment when contacted by WIRED to confirm whether such activity actually occurred on the personal Star Wars order. Despite the chatbot’s automated compliance, internal safeguards within Lego’s AI assistant could have flagged the unusual purchase volume, potentially triggering human review.
The Vulnerabilities of Agentic Commerce
Musk claimed users could let Grok buy items by uploading a picture of their credit card, allowing the bot to search the web for deals and complete transactions. Major AI developers, including Meta and OpenAI, have similarly promoted their respective agents for shopping tasks.
This automated capability introduces significant financial risks. At the end of 2025, Visa issued a warning regarding fraud in agentic commerce. The payment processor advised that malicious actors were building legitimate-looking sites designed to trick AI agents into purchasing fake items through artificially low pricing.
“Once the AI agent completes the purchase … the malicious merchant can harvest payment data and instantly use it for unauthorized transactions,” according to Visa’s blog. Visa documented a 25 percent increase in malicious bot-initiated transactions during the latter half of 2025, with the figure spiking to 40 percent inside the United States.
