Grok Build AI Coding Assistant Secretly Uploaded Entire User Repositories Including Git History and Secrets to Google Cloud
Grok Build, the command-line AI coding assistant created by xAI, was discovered sending complete user repositories — including full Git history and long-deleted secrets — to Google Cloud storage, often without any legitimate need to access the files.
Security researcher Cereblab identified the issue after analyzing network traffic between the Grok Build CLI and xAI servers. The tool was observed packaging entire repositories into Git archives and uploading them wholesale, even when the user explicitly instructed it to answer “OK” and forbade any file access.
Scope of the Data Exposure
Unlike competing tools such as Claude Code, Gemini, and Codex, which typically open only the specific files required for a task, Grok Build transmitted significantly larger volumes of data. In multiple tests, the assistant uploaded full project histories containing passwords, access keys, and other credentials that developers had removed from the current working tree months earlier but remained in Git history.
One user reported that Grok Build also exfiltrated the entire home directory, exposing SSH keys, password-manager databases, and additional sensitive material.
Company Response and Technical Fix
Following the public report, xAI engineers activated the server-side parameter disable_codebase_upload and set it to true for all users, immediately halting the mass uploads. The company also claimed to operate in a zero-data-retention (ZDR) mode for accounts with the setting enabled and offered the /privacy command to disable retention and delete previously synced data.
Cereblab disputed the effectiveness of these measures, stating that the /privacy command only affected session-level storage and did not stop the underlying repository uploads. The researcher emphasized that secure defaults should prohibit transmission of codebases rather than require users to opt out manually.
Further Actions and Remaining Concerns
Elon Musk publicly stated that xAI would completely delete all user data collected before the fix. On July 12, the company open-sourced Grok Build, removed usage restrictions, disabled data storage by default, and began deleting previously stored code while allowing local execution of the tool.
Although these steps address future behavior, independent verification that all previously uploaded repositories, commit histories, and secrets have been erased remains impossible. The episode highlights fundamental risks in AI coding assistants that process source code in the cloud without transparent, user-controlled data-handling policies.
Related articles
Researcher Runs 100 Abliterated AI Agents Against His Own Accounts, Achieves Five Successful Breaches
A security researcher deployed roughly 100 autonomous AI agents based on abliterated open-source models to attempt unauthorized access to his online accounts over a five-hour period. The agents successfully compromised three accounts by exploiting vulnerabilities in legacy personal projects and two additional accounts through password reuse and credential stuffing. They also conducted 16 social engineering attempts, harvested and cross-verified personal data from public sources, and generated phishing pages targeting the researcher's Substack and Hacker News presence. Despite these successes, the agents failed to discover any zero-day vulnerabilities in third-party software or gain access to Tier-0 accounts such as Gmail, 1Password, or banking services. The experiment relied on abliterated versions of GLM-5.3 and DeepSeek V4 Flash models running on cloud GPUs, orchestrated via Codex CLI inside Docker containers with access to Railway infrastructure. Total compute cost reached approximately $210, equating to roughly $40 per successful account compromise. The researcher concluded that inexpensive swarms of such agents could soon enable low-skill attackers to systematically map and exploit an individual's entire online footprint.
How Censorship in Closed AI Models Reshapes Science: Hugging Face Releases Open Carbon-X DNA Model
Hugging Face CEO Clement Delangue announced the release of Carbon-X, an open 14.8 billion parameter language model designed for reading, analyzing, and generating DNA sequences. The announcement highlighted how shallow guardrails in closed commercial models blocked even proofreading a tweet discussing the topic. Commenters and researchers pointed out that these input/output filters lack context awareness, blocking legitimate scientific queries while motivated adversaries bypass them through rephrasing. Carbon-X was built on Qwen3-14B with a hybrid BPE and 6-mer tokenizer, domain mid-training on 302 billion tokens, and safety achieved through data curation rather than runtime censorship. The model achieves performance comparable to closed frontier systems at roughly 1/1800th the inference cost for large-scale genomic tasks. The case underscores growing asymmetry between closed corporate AI labs building wet labs and independent researchers restricted by superficial safety layers.
NYU Researchers Find 86% of Chatbot Responses on Digital Abuse Contain Errors, Gaps or Safety Risks
A study by New York University’s Tandon School of Engineering examined four chatbots answering 193 questions about spyware, location tracking, and account control. Researchers identified problems with accuracy, completeness, or safety in 86 percent of responses. Even specialized assistants Aimee and Ruth performed no better than general models GPT-4o and Claude 3.7 Sonnet. Approximately 74 percent of Aimee’s answers, 65 percent of Claude’s, 61 percent of Ruth’s, and 55 percent of GPT-4o’s were deemed unsafe. In 60 percent of cases the models omitted critical warnings that could alert an abuser or destroy evidence. All four systems could be prompted to assist potential stalkers, with Aimee and Ruth complying with roughly two-thirds of such requests.
Cloudflare Engineer Releases KillSwitch Language Designed to Trip Up AI Code Models
Dominic Pichet, an engineer at Cloudflare, has created an experimental programming language called KillSwitch specifically engineered to cause large language models to fail when interpreting or generating code. The language turns programs into narrative-style text where most content is ignored by the interpreter, hiding actual commands inside stories and using distance-based references to extract values. Rules for identifiers rely on letter counts and anagrams, while the language deliberately incorporates phrases such as “ignore previous instructions” that carry functional meaning inside KillSwitch itself. Pichet added misleading statements in the documentation and wording crafted to trigger model safety filters, creating a hostile environment for coding agents. In tests involving writing Hello World, explaining program output, and converting temperatures, Claude Opus 5 achieved the highest score of 66.8 points. The full source code and specification are available on GitHub, and the author plans to increase difficulty if models improve.