AI hardware theft: when even GPS can lie
More than $100 million diverted since June
An investigation published on October 26 describes networks believed to have stolen more than $100 million in servers, memory, fibre, copper and other equipment bound for US data centres. In two of the largest cases, shipments valued at $13 million and $10 million disappeared after vehicles deliberately struck their security escorts. In the second case, the tracking application kept showing apparently normal coordinates because the location had reportedly been spoofed. Read the Ars Technica investigation
A map position does not prove possession
The attack begins before the road: broker accounts can be compromised, fake postings lure legitimate carriers, and manipulated documents or driving licences place the wrong driver at the right dock. If an operation trusts only the tractor GPS or an application controlled through a compromised account, the dashboard may confirm a journey that no longer corresponds to the physical cargo.
The cargo unit needs its own identity
Independent sensors on the pallet, container or equipment can combine location, opening, motion, shock and loss-of-signal data. Every reading should be bound to the driver, vehicle, carrier and authorisation valid at that moment. A blockchain record preserves dispatches, receipts and exceptions across companies, making it harder to rewrite the operating history after a diversion appears. Explore Sensefinity blockchain
The passport should record custody, not only origin
A Digital Product Passport can connect each server, module, spool or pallet to its origin, serial number, authorised carrier, seals and expected handoffs. A pickup by a newly controlled business, an unexpected plate or a driver change becomes a verifiable exception before the goods leave, rather than an explanation reconstructed after the loss. Explore Sensefinity Digital Product Passport
Recovery does not prove that cargo is intact
After recovery, operators must validate identity, packaging, shock, humidity, temperature and possible substitution before stock is reinstated. The discipline FoodSteps applies in food chains — connecting condition, lot, inspection and disposition — is a useful model for any sensitive shipment: finding it is only the beginning; release requires evidence of its state. Visit FoodSteps
- Confirm current carrier control and driver identity before pickup.
- Bind the sensor to the pallet or equipment, not only the tractor.
- Compare GPS with independent opening, motion, network and physical milestones.
- Block critical route, plate or driver changes without renewed authorisation.
- Preserve exception and inspection events in a shareable chain of custody.