A Grok Vibe-Coding Agent Leaked Sensitive SpaceXSI Data

>_ mC0de INSIGHTS / mC0de CyberSecurityINSIGHTS ONLINE
CYBERSECURITY

A Grok Vibe-Coding Agent Leaked Sensitive SpaceXSI Data

A design flaw in the Grok Build vibe-coding agent allowed sensitive data from thousands of users, including SpaceXSI itself, to be exposed publicly. Security researchers discovered that improper URL structuring let passive DNS services enumerate and capture live projects.

November 11, 2026  ·  mC0de CyberSecurity

A major privacy exposure has hit thousands of users utilizing SpaceXSI's Grok Build vibe-coding agent. Security firm Claroc discovered that a structural flaw in how the platform handles sharing links allowed sensitive corporate secrets, employee records, and internal plans to be leaked publicly. The vulnerability stemmed from placing randomly generated dice words to the left of the domain rather than the right. This implementation error made the URLs visible to passive DNS services and publicly searchable databases. While SpaceXSI secured its own exposed apps after public notification, thousands of other user projects remained vulnerable.

24Exposed SpaceXSI Apps
38 bitsLink Entropy Level
2 percentMaterial Rated High or Critical
01 / WHAT HAPPENED

The Core Vulnerability in Grok Build

Security firm Claroc revealed that a design decision by developers of the Grok platform completely undermined its security objectives. The platform utilizes link-sharing features so users can quickly share apps created through text or speech inputs with authorized individuals.

However, the platform positioned random dice words to the left of the domain instead of the right. This architectural choice allowed passive DNS servers to capture lookup requests in publicly searchable databases, exposing sensitive files to anyone.

Dice Words

Randomly generated phrases used in URLs to provide entropy, intended to prevent unauthorized guessing.

Passive DNS

Services monitoring central DNS lookup servers that inadvertently captured and indexed the unmasked URLs.

02 / THE IMPACT

Exposed Data Across Multiple Organizations

The exposed information included proprietary business plans, medical examination results, ad proposals, and internal system addresses. Affected entities ranged from a Peruvian construction company and a luxury cruise line to major Fortune 500 organizations.

Specific leaks involved employee records, security training matrices, sales territory maps, and logistics support contracts. Security analysts estimated that roughly two percent of the discovered material held enough sensitive data to be rated as high or critical severity.

Corporate Leaks

Exposed internal sales territory maps, advertising proposals, and onboarding plans belonging to various companies.

Employee Data

Medical exam results, worker rosters, and security clearance documents leaked via unauthenticated app links.

03 / HOW IT WORKS

Contrasting Architectural Implementations

Capability-based access control relies on possession of a secure URL serving as authorization to resources behind it. Platforms like Google successfully use this method by placing random identifiers to the right of the domain, keeping queries hidden from central tracking.

In contrast, SpaceXSI placed the random phrase on the left side of the domain. This fundamental divergence allowed third-party monitoring services to easily enumerate live projects, turning private workspace links into public endpoints.

04 / INDUSTRY RESPONSE

Response Times and Remediation Efforts

Claroc notified SpaceXSI of the privacy spill, but received no immediate response until security officials publicly tagged executives on X. Following public pressure, SpaceXSI removed twenty-four of its own exposed applications within hours.

Despite these swift actions regarding internal tools, many apps created by external organizations and other Grok users remained publicly accessible. Security experts and affected companies continue working to contain the ongoing exposure across various industries.

THE mC0de TAKEAWAY

Why It Matters

The recent Grok Build security incident highlights the critical risks associated with rapid vibe-coding and AI-driven application deployment. While generative tools significantly accelerate the timeline for building internal software, they often bypass standard enterprise security reviews and authentication guardrails. Developers and platform architects must ensure that capability-based access controls strictly adhere to established privacy patterns, such as placing entropy to the right of domains. Otherwise, the convenience of instant link-sharing can easily transform into an expansive attack surface for corporate espionage and accidental data leakage.

Research & Sources

A Grok vibe-coding agent leaked sensitive SpaceXSI data (engadget.com) · Grok Build coding agent leaks oodles of confidential info, some belonging to SpaceXSI (arstechnica.com)

Reporting reviewed through November 11, 2026.

← Explore More Insights