Grok Build default privacy settings expose thousands of application data stores
Security researchers found that default public sharing settings in SpaceX Super Intelligence's Grok Build agent exposed more than 35,000 application environments and confidential data.
FTMQ SI, written by our newsroom0 views

A security researcher discovered that more than 35,000 Grok Build applications were publicly accessible online, Ars Technica AI and MLQ.ai reported. Security researcher Aprozper conducted a passive domain name system search that returned 43,601 links ending in the domain grok.me, according to Ars Technica AI. Of those total discovered links, nearly 95 percent were set to public, allowing anyone who clicked on a link to access the corresponding application and its stored data. [1][6]
The vulnerability stems from default access configurations in the platform. Ars Technica AI reported that tests conducted over a one week period showed Grok Build sets application links to be available to anyone with the link by default. Users retain the option to alter these settings to restrict access to invited users or to only the creator. Confidential material belonging to SpaceX Super Intelligence, including sales data, was exposed among the accessible files, according to Mezha.net and MLQ.ai. [1][3][6]
A representative for Cursor, an artificial intelligence coding service recently acquired by SpaceX Super Intelligence, disputed the findings in an email cited by Ars Technica AI. The representative stated that Grok Build applications are private by default and nothing becomes public unless a creator actively chooses to share it. The representative added that competing coding agents from ChatGPT and Lova operate under different sharing configurations. [1]
Grok is a family of generative artificial intelligence models developed by SpaceX Super Intelligence, an enterprise founded by Elon Musk that became a subsidiary of SpaceX in February 2026 [7, 8]. The company operates the Cursor software development environment, and its Grok Build tool allows developers to execute an agent inside codebases using interactive terminal interfaces or scripts [8, 9]. The Grok Build platform runs on the Grok 4.7 model architecture and supports sandboxed command execution alongside model context protocol servers [9]. [7][8][9]
In short
- A passive DNS search identified 43,601 grok.me links, of which over 35,000 were publicly accessible.
- Tests indicated Grok Build application links defaulted to shareable with anyone who had the link.
- Exposed data included confidential SpaceX Super Intelligence sales material.
- Cursor representatives claimed that Grok Build applications are private by default unless shared by the creator.
Sources
Every paragraph above points to the numbered items it rests on. Read the originals here.
- [1]Grok Build coding agent leaks oodles of confidential info, some belonging to SpaceXSIArs Technica AI, 7d ago (the report this story comes from)
- [2]Grok Build coding agent leaks oodles of confidential info, some belonging to SpaceXSIArs Technica, 7d ago
- [3]Grok Build exposed confidential data, including SpaceXSI materialMezha.net, 7d ago
- [4]Sinch joins SpaceXSI network to aid Grok Build appsIT Brief UK, 9d ago
- [5]Grok Build Sent 5.1 GiB for a 192 KB Task [2026]shattered.io, 7d ago
- [6]Grok Build left 35,000 apps open to anyone, including SpaceX's own confidential sales materialMLQ.ai, 7d ago
Background
- [7]Grok (chatbot) on Wikipedia
- [8]SpaceXAI on Wikipedia
- [9]Grok Build on Grokipedia
Our newsroom writes these reports with the help of software, from the 9 sources listed and nothing else, and checks them against those sources. Facts can still be wrong or move on; the originals are the record. Spotted a mistake? Write to daniel@monsterkong.com.
Related from FTMQ SI
Earlier reports of ours on the same people and subjects.
- Rogue AI agents accessed personal data at Indian organizationsTop Stories, 32d ago
- Hundreds of exposed Artex AI hacking tools discovered onlineTop Stories, 11d ago
- California sues OpenAI following autonomous agent safety breachesTop Stories, 1d ago
- ServiceNow unveils AI Control Tower platform for enterprise agent governanceModels & Products, 31d ago
- Elon Musk exposes order data after posting screenshot of Grok purchaseTop Stories, 36d ago
More in Top Stories
- Meta executive posts AI video showing deepfaked tech leaders4h ago
- Amazon and Google to control combined 30 percent Anthropic stake post IPO4h ago
- Nvidia ends H200 sales to China following leadership changes4h ago
- Nvidia board rejects calls to reinstate former chief executive Jensen Huang1d ago
- Donald Trump defends Jensen Huang following ouster over alleged spying affair1d ago
- California sues OpenAI following autonomous agent safety breaches1d ago
Get the day in one email
Reports like this one, the top news of the last 24 hours, every morning. Free, one email a day; readers can comment under every report.
By signing up you agree to our terms and privacy policy. Unsubscribe any time.

Comments
Loading
Join the conversation
Comments are open to readers of our daily email: the top news of the last 24 hours, every morning, free. Sign up and the comment box opens.
Already on the list? Enter the same address and we will send a sign-in link.
By signing up you agree to our terms and privacy policy. Unsubscribe any time.