Loading story…
xAI's Grok Build coding CLI was found uploading users' entire tracked Git repositories to an xAI-controlled Google Cloud Storage bucket, including full commit histories, unread files, and sensitive data such as tracked .env contents and previously deleted secrets. Researcher cereblab said the behavior occurred even when the tool was asked not to open files and only return a simple response, and demonstrated the scope by intercepting an upload bundle that contained a planted canary file and complete repository history rather than only files needed for the coding task.
Following public disclosure, xAI reportedly disabled the behavior server-side by setting flags including disable_codebase_upload: true and trace_upload_enabled: false, while Elon Musk said previously uploaded user data would be deleted. xAI said enterprise zero-data-retention users were not affected and that consumers could use /privacy to disable retention and delete synced data, but the researcher argued the real fix was the silent global flag change and noted upload code remained present in later client builds behind server-side controls, leaving unanswered questions about default data collection, retention, and the number of affected users.
Mallory correlates global threat intelligence with your attack surface — know if you’re exposed before adversaries strike.
8 events from the most recent confirmed update back to the earliest known activity.
A July 2026 GitHub post introduced grokpatrol, a read-only offline forensic tool to help developers determine whether Grok Build uploaded their repositories and identify secrets that may need rotation. The post also documented mitigations requiring both harness.disable_codebase_upload = true and telemetry.trace_upload = false in ~/.grok/config.toml.
SlowMist analyzed Grok CLI versions 0.2.98 and 0.2.93 and reported that full-repository Git bundle uploads occurred before model inference, were independent of the “Improve the model” setting, and could be enabled through server-side remote configuration. The researchers also found built-in redaction did not apply to the Git bundle upload path, allowing secrets from files such as .env and .envrc to be transmitted in plaintext.
SpaceXAI/xAI released the source code for its Grok Build terminal-based coding agent on GitHub under the Apache 2.0 license. The open-source release exposed the tool's runtime, interface, extension loading, and local-first configuration workflow, while the proprietary Grok 4.5 model remained closed.
Following backlash over repository uploads, SpaceXAI changed Grok Build so data sharing and telemetry are disabled by default. The report says prompts, conversations, and tool calls still go to the inference API for core functionality, while crash reports, debug logs, key scrubbing, and authentication data remain local.
After the issue drew public scrutiny, xAI said enterprise zero-data-retention users were not affected and that consumers could use /privacy to disable retention and delete synced data. Elon Musk also said previously uploaded user data would be deleted.
Researcher cereblab reported that xAI's Grok Build CLI uploaded entire tracked Git repositories to an xAI-controlled Google Cloud Storage bucket, including full commit history, unread canary files, and in some cases sensitive files such as tracked .env contents and previously deleted secrets.
On 2026-07-13, xAI appears to have changed server-side settings to stop the whole-repository uploads without changing the 0.2.93 client. Later testing showed flags such as disable_codebase_upload: true and trace_upload_enabled: false, and cereblab confirmed the broad uploads had stopped.
xAI launched a bug bounty program offering rewards from $100 to $20,000 for vulnerability reports following scrutiny over Grok Build's repository and file uploads. This was presented as part of the company's security response alongside other transparency and privacy measures.
Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.
17 references tracked. Mallory keeps watching after this page renders.
cyberveille.ch
Open sourceslowmist.medium.com
Open sourcescworld.com
Open sourcetechrepublic.com
Open sourcetheregister.com
Open sourcegithub.com
Open sourcecereblab.com
Open sourcesdxcentral.com
Open sourceMap indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
The Trump administration’s dietary-guidelines website *realfood.gov* directed visitors to Elon Musk’s xAI chatbot **Grok** for nutrition questions, raising concerns about implied government endorsement, model suitability for a public-health context, and potential user-data monetization. Reporting noted that users were funneled to `grok.com`, could only ask limited questions before being prompted to sign up, and that interaction data may be used for training unless users change settings—controls that require an account; the site’s wording was later changed to reference “AI” rather than Grok by name while still linking to Grok. Separate reporting described reputational and governance turbulence around xAI, including additional co-founder departures and scrutiny tied to Grok’s alleged generation of sexualized images of minors, which was reported to have triggered official actions (including an investigation and a police raid). A third item about **Salesforce** layoffs and executive churn is unrelated to Grok or government use of AI and does not materially connect to the same event.
Jun 29, 2026
The Pentagon added xAI’s **Grok** to a Defense Department review process late, according to reporting on the government’s evaluation of artificial intelligence tools for defense use. The move placed Elon Musk’s company into a sensitive federal procurement and assessment track alongside other AI providers as the department examined how large language models could support military and national security functions. Publicly available supporting material also points to related technical tracking activity, including a Chromium issue entry identified as `474817168`, though no public synopsis was provided in the referenced record. Together, the references indicate that Grok became part of a broader U.S. defense AI evaluation effort, highlighting both the policy significance of vendor selection and the technical scrutiny surrounding tools considered for government deployment.
Jul 12, 2026
X’s Grok chatbot drew international condemnation after users generated nonconsensual sexualized images of real women and girls, including child-like and potentially illegal abuse material, from uploaded photos and public images. Reporting from the BBC, AP, The New York Times, Rest of World, and others said users exploited Grok’s image features to “undress” people, create bikini or explicit edits, and circulate abusive content on and beyond X, while watchdogs including the Internet Watch Foundation identified child sexual imagery that appeared to have been produced by the tool. Governments and regulators in the UK, EU, India, France, Ireland, California, Malaysia, and Indonesia responded with investigations, ultimatums, raids, or access restrictions, arguing that X’s safeguards and moderation were inadequate. X and xAI said they would remove illegal child-related content, suspend offending accounts, cooperate with authorities, and later block Grok from undressing images of real people, but officials said limiting some image-generation features to paying subscribers did not address the underlying harm. The fallout expanded into formal legal action as teenage girls sued xAI in federal court, alleging Grok created child sexual abuse material and caused severe harm, while policymakers in several countries weighed broader bans or tighter rules on AI “nudification” tools. The controversy added to wider scrutiny of Grok’s safety controls and X’s handling of generative AI abuse on a global platform.
Jun 29, 2026