ThisCalculator

Everyday

Verify AI Server Provenance & Chain of Custody

Updated · Standard formula, shown below

This guide explains how to verify the provenance and chain of custody of AI servers before deployment, ensuring the hardware is authentic, untampered, and has a documented history from manufacturing to your data center. This process reduces security risks, prevents counterfeit components, and supports compliance audits.

What You’ll Need

  • Server serial numbers and model numbers (from the purchase order and shipping manifest)
  • Manufacturer-provided certificate of authenticity or tamper-evident seals (if applicable)
  • Access to the manufacturer’s verification portal or API (or a phone number for manual verification)
  • Cryptographic hash tool (e.g., sha256sum, Get-FileHash) for firmware/software verification
  • Secure logging system (e.g., a spreadsheet, database, or dedicated chain of custody tool) to record each step
  • Two-person integrity for critical steps (optional but recommended)

Step-by-Step

  1. Document the expected chain. Before the server arrives, record the expected path: manufacturer → distributor → reseller → your organization. Note the expected serial numbers, part numbers, and any order or shipping references from your purchase documentation.
  2. Inspect the physical packaging upon delivery. Check for tamper-evident tape, security seals, or shrink wrap. Look for signs of re-sealing, cuts, or mismatched logos. Photograph the packaging from all angles before opening.
  3. Verify the serial number and model. Compare the serial number on the server chassis, on the box, and on the invoice. Use the manufacturer’s official verification tool (website, API, or phone) to confirm the serial number is valid and matches the model you ordered. Record the verification result with a timestamp.
  4. Inspect the internal hardware. Open the chassis and visually inspect for any signs of tampering: loose screws, non-original components, solder marks, or missing tamper-evident stickers on internal parts (e.g., GPUs, RAM, motherboard). Use a flashlight and magnifier if needed.
  5. Verify component serial numbers. For critical components like GPUs, CPUs, and storage drives, record their serial numbers and check them against the manufacturer’s database if available. This helps detect swapped or counterfeit parts.
  6. Check firmware and software integrity. Boot the server into a trusted environment (e.g., a live Linux USB) and compute cryptographic hashes of the firmware and pre-installed software against the manufacturer’s published hashes. Use a command like sha256sum on the firmware image or verify the secure boot signature.
  7. Confirm secure boot and TPM status. Enter the BIOS/UEFI and verify that Secure Boot is enabled and that the TPM (Trusted Platform Module) is present and functional. Record the TPM endorsement key hash if possible.
  8. Test hardware functionality. Run a hardware diagnostic tool (e.g., memtest86, stress-ng) to ensure all components function correctly. This catches failures that might indicate refurbished or damaged parts.
  9. Record the chain of custody. Log every step with the date, time, person responsible, and any evidence (photos, hash values, verification results). Store this log in a secure, tamper-evident system.
  10. Seal and secure the server. After verification, apply your own tamper-evident seals (e.g., security screws, tamper-proof labels) to the chassis to prevent unauthorized access during deployment. Record the seal numbers.

Common Mistakes

  • Skipping the visual inspection. Relying solely on serial number verification can miss physical tampering.
  • Not verifying the manufacturer’s verification tool. Always use the official manufacturer channel; a third-party site might be a phishing attempt.
  • Ignoring the TPM. A missing or non-functional TPM can compromise attestation and secure boot.
  • Recording hashes without context. A hash is only useful if you know what file it belongs to and the expected value.
  • Failing to photograph the unboxing. Photos provide visual evidence that can be used in disputes.

Tips

  • Perform the verification in a secure, access-controlled area to prevent tampering during the process.
  • Use a two-person rule: one person performs the inspection, another records the results, to reduce the risk of internal fraud.
  • Keep the original packaging and seals until the server is fully deployed and tested.
  • For high-security environments, consider using a hardware security module (HSM) to store the chain of custody log.
  • If the manufacturer offers a blockchain-based provenance service, use it if available—it provides an immutable record.

Troubleshooting

  • Serial number not found in manufacturer’s database. Double-check the number for typos. If still not found, contact the manufacturer directly with photos of the chassis. Do not deploy until resolved.
  • Tamper-evident seal appears broken. Do not accept the server. Photograph the seal and refuse delivery, or if already accepted, quarantine the server and contact the supplier immediately.
  • Firmware hash does not match. This could indicate a compromised or unofficial firmware. Do not boot the server into your network. Contact the manufacturer for guidance; you may need to reflash the firmware with a trusted image.
  • TPM is not detected. Check BIOS settings to ensure TPM is enabled. If it’s still not detected, the TPM may be faulty or removed—contact the manufacturer for a replacement.
  • Hardware diagnostic fails. A failed component might be defective, not necessarily tampered. Document the failure and request a replacement or RMA. If the failure is suspicious (e.g., a GPU with a different serial number), treat it as a security incident.

Frequently Asked Questions

What exactly is chain of custody for servers?

Chain of custody is a documented, unbroken trail of who handled the server, when, and where, from the point of manufacture to your deployment. It ensures that no unauthorized party had access to the hardware.

Is this process required by law?

Not generally, but it may be required by industry regulations (e.g., healthcare, finance) or for compliance with security frameworks like SOC 2 or ISO 27001. Always check your organization’s requirements.

Can I trust a third-party verification service?

Only if it is officially endorsed by the manufacturer. Otherwise, the service could be a scam. Always use the manufacturer’s own verification channel.

What if I can’t find the serial number on the manufacturer’s website?

Contact the manufacturer’s support line. They can help verify if the number is valid. If they cannot confirm, treat the server as potentially counterfeit and do not deploy.

How often should I re-verify the chain of custody?

For most environments, a one-time verification at deployment is enough. For high-security setups, you may re-verify after any hardware maintenance or relocation.

Do I need to verify every component’s serial number?

Not necessarily. Focus on high-value or security-critical components like GPUs, CPUs, and storage. For others, a visual inspection may suffice.

Related

Method: the standard formula for this calculation, rounded as each result states. See how the calculators work or report an error.