Slide deck
BackGrok Build Shipped Whole Repos to xAI's Cloud
A wire-level teardown found Grok Build bundling whole git repositories, history included, and uploading them regardless of what the agent read — 5.10 GiB against 192 KB of model traffic. xAI (now SpaceXAI) disabled it server-side within a day; per cereblab, no user setting had.
AI Security · Wire-Level Teardown
Grok Build Shipped Whole Repos to xAI's Cloud
A wire-level teardown found Grok Build bundling whole git repositories, history included, and uploading them regardless of what the agent read — 5.10 GiB against 192 KB of model traffic.
Grok Build · wire-level teardown1 / 20
What Happened
By the Numbers
one 12 GB repo, one session
- 27,800×more data left the machine than the model's turns needed
- 5.10 GiBuploaded via /v1/storage — against 192 KB of model traffic
- 0settings found that stopped the upload, before the 13 Jul fix — “I did not find one”
- 2unrelated repositories — a never-read canary recovered from each
Grok Build · wire-level teardown2 / 20
What Was Proven — and What Was Not
Confirmed
What Who says so At launch, xAI marketed Grok Build as “local-first,” stating no source code was transmitted to its servers DevOps.com's launch coverage, 15 May 2026, read directly On 13 July the whole-repository upload stopped; the server began returning disable_codebase_upload: true— confirmed independently on a second accountcereblab's own retest; developer Peter Dedene's own account, posted on X Grok Build · wire-level teardown3 / 20
What Was Proven — and What Was Not
Confirmed
What Who says so xAI (renamed SpaceXAI on 6 July) publicly acknowledged that data retention had been on by default for non-enterprise users, and Musk pledged to delete all previously-uploaded data SpaceXAI's own X post; The Register, 14 Jul 2026 SpaceXAI open-sourced the Grok Build CLI under the Apache License on 16 July; the upload code was found still present in the release, altered rather than removed The Register, 16 Jul 2026, citing developer Simon Willison's own review Grok Build · wire-level teardown4 / 20
What Was Proven — and What Was Not
Confirmed, but not verifiable here
cereblab's wire capture, with a public reproduction harness — not independently reproduced by a named third party in the sources below
What Grok Build 0.2.93 ran a second channel that bundled the whole repository, full git history included, and uploaded it via /v1/storageregardless of whether the agent had been told to read anything; on a 12 GB test repo, that channel moved 5.10 GiB against 192 KB of model-turn trafficCloning a captured bundle recovered a file the agent had explicitly been told not to open, verbatim, plus full git history; replicated on a second, unrelated repository A tracked .envfile's secrets were transmitted unredacted, both in the model-turn traffic and in the storage uploadTurning off “Improve the model” did not stop the bundle upload; the server continued to report trace_upload_enabled: trueGrok Build · wire-level teardown5 / 20
What Was Proven — and What Was Not
Unconfirmed · Not public
How firm What Who says so Unconfirmed That the pledged deletion of previously-uploaded data was actually completed Not found in any source below, as of 25 Aug 2026; The Register's own 14 Jul report says the same about its own coverage Unconfirmed That “other researchers” independently confirmed the finding on their own repositories, and that Codex and Anthropic “independently found indirect evidence” of eight further private repositories fully uploaded cereblab's own site (unnamed, no linked artifact); developer Peter Dedene's own account of what he says those parties found — neither corroborated by a statement from OpenAI or Anthropic Not public Whether SpaceXAI trains its models on data collected this way cereblab's own gist states plainly this was not proven either way: “upload/storage ≠ training” Grok Build · wire-level teardown6 / 20
Timeline
How It Unfolded
eleven days from teardown to retest, then a five-week silence on completion
- 2026-02-02SpaceX completes its all-stock acquisition of xAI.
- 2026-05-15Grok Build is covered in early testing, marketed as “local-first” — trade press reports xAI's own claim that no source code is transmitted to its servers.
- 2026-07-06xAI's public accounts rebrand as SpaceXAI.
- 2026-07-12cereblab publishes the wire-level teardown of grok build 0.2.93, with SHA-256-hashed evidence and a public reproduction harness.
Grok Build · wire-level teardown7 / 20
Timeline
After the Teardown
- 2026-07-13The upload stops: the server begins returning
disable_codebase_upload: true. Developer Peter Dedene reports the same flag on his own account. - 2026-07-14The Register and The Hacker News report the findings. Musk publicly pledges to delete all previously-uploaded user data — not yet independently confirmed as complete.
- 2026-07-16SpaceXAI open-sources the Grok Build CLI under the Apache License and resets usage limits. Independent developer Simon Willison reviews the released source: the upload code is still there, altered to reverse the behaviour rather than removed.
- 2026-07-22cereblab retests grok build 0.2.106 and reconfirms the fix holds.
Grok Build · wire-level teardown8 / 20
- 2026-07-13The upload stops: the server begins returning
The Argument
Two Channels, Different Rules
the permission prompt sees only one
Channel A · read-driven
/v1/responses
the model's turns
- Carries only what the agent actually opened.
- This is the channel the permission prompt governs.
- Never-read files are genuinely absent here.
Channel B · repo-driven
/v1/storage
the git bundle
- Bundles git-tracked files plus full history.
- Indifferent to what the model read — or didn't.
- 5.10 GiB from the same session; 73 chunks, all HTTP 200.
- Destination: a Google Cloud Storage bucket.
Grok Build · wire-level teardown9 / 20
The Argument
The Canary That Never Was Read
The prompt: “Reply with exactly: OK. Do not read or open any files.” The agent opened nothing — and POST /v1/storage still returned 200.
- OKthe agent's entire reply — it opened nothing
- 4commits recovered from the bundle, in full
- ×2codebases the recovery was repeated on
Confirmed, but not verifiable here
The harness that does this is public, with its method published in full — the account above is cereblab's own, and is not known to have been independently re-run by anyone else.
Grok Build · wire-level teardown10 / 20
The Argument
On the Wire
one session · two destinations
What went up Channel Evidence The repository, as a git bundle /v1/storage 73 chunks of ~75 MB, every one HTTP 200; cloned back to recover a never-read canary. A .env, unredacted both A tracked .envfile's secrets were transmitted unredacted, both in the model-turn traffic and in the storage uploadDestination bucket GCS Named in the binary's strings and in staged metadata paths. Grok Build · wire-level teardown11 / 20
Three Switches, None of Them the Brake
Control What a developer assumes What it actually did “Improve the model” toggle · default ON opts me out of data collection governed training/retention policy, not transmission. Switched off, the server still returned trace_upload_enabled: true and the bundle still uploaded — until the 13 Jul server-side fix. /privacy command (added after the fix) stops the sending a data-retention setting — “not a block on what's sent”, wire-tested by cereblab. What actually stopped transmission was the separate disable_codebase_uploadserver flag.Any user-side off switch there must be one somewhere none was found before 13 Jul. The upload stopped only when SpaceXAI flipped a flag on its own servers — a remedy a user could not reach. Grok Build · wire-level teardown12 / 20
The Argument
The Failure Is in the Consent Model
Bottom line
The failure is in the consent model, not any single line of code. The prompt asked “may I read?” while a second channel bundled git-tracked files plus full history and shipped them regardless — and the only thing that ever stopped it was the vendor, on its own servers.
Reframe
Your exposure is whatever the tool uploads — never whatever it reads.
Grok Build · wire-level teardown13 / 20
What Others Add
Why a Deleted Secret Still Shipped
the exposure chain
- 01Commita secret, months ago
- 02Deletefrom the working tree
- 03Persistit lives on in git history
- 04Bundletracked files + history
- 05Uploadregardless of reads
- 06Rotatethe only real remedy
Grok Build · wire-level teardown14 / 20
What Others Add
How the Field Compares
cereblab's own cross-tool tests
Tool Uploads the whole repository? Standing Claude Code No — only files it opens Confirmed, but not verifiable here Codex No — only files it opens Confirmed, but not verifiable here Gemini No — only files it opens Confirmed, but not verifiable here Grok Build (0.2.93, before the fix) Yes — whole repo plus git history, to the cloud Confirmed, but not verifiable here Grok Build · wire-level teardown15 / 20
What Others Add
The Company's Own Words
Both statements are given here side by side, neither resolved in favour of the other.
Its account
“Since launch, Grok Build has fully respected zero data retention (ZDR)… In the early beta, data retention was enabled by default for non-ZDR users. Based on your feedback, we changed this.”
Its denial
SpaceXAI says “all users have always had the ability to disable data upload in the CLI” and that “when data upload was disabled, this choice was respected” — a flat denial of the claim that no user-side setting stopped the bundle.
Grok Build · wire-level teardown16 / 20
What Others Add
What Followed
The date it gives
It also dates its own change to “starting on July 12th,” a day before the server flag was observed to change.
What it has not published
What SpaceXAI has not done, in any source below, is publish a security advisory, a changelog entry, or a CVE for the behaviour — its account of events lives in social posts.
Grok Build · wire-level teardown17 / 20
Conclusion
Recommendations
highest-return first
- Rotate every credential reachable in tracked git history — deleting it never removed it, and a pledge to erase uploaded data cannot un-ship what already left the machine.
- Verify on the wire, not in the prompt — a permission dialog describes one channel; measure what actually leaves the machine, the way a public reproduction harness lets a third party do.
Grok Build · wire-level teardown18 / 20
Conclusion
Recommendations, continued
- Demand default-off — a retention toggle you must run every session is not consent, and it is a different control from the one that actually stops transmission. The default is the policy.
- Treat a vendor-side fix as a pause, not a close-out — the upload code itself was reported still present, merely altered, after the fix; and whether previously-uploaded data was actually deleted remains unconfirmed. Prefer tools whose off switch you hold locally and can verify.
Grok Build · wire-level teardown19 / 20
Conclusion
The Right Default Is Off
A remedy you cannot reach is not a control.
Where it lands
Any agentic coding tool whose consent surface is scoped to a narrower channel than its transmission has this bug — whatever its logo.
Still open
Whether the pledged deletion was completed remains unconfirmed. No CVE identifier or formal vendor security advisory for this behaviour appears in any source below, as of 25 Aug 2026.
Grok Build · wire-level teardown20 / 20