Cheap, fast and the size of a briefcase.
“Now that we have redistributed DDR5 traffic, our work shows that even the most advanced TEEs from all manufacturers with available hardware are vulnerable to low-cost physical attacks,” Genkin said.
The equipment required for TEE.fail includes standard equipment costing less than $1,000. One of the devices the researchers created fits into a 17-inch briefcase so it can be smuggled into a room containing a TEE-protected server. Once a physical attack has been carried out, the device no longer needs to be connected. Attackers who hack TEE on servers they control do not need stealth, allowing them to use larger devices that the researchers have also created.
Mechanism connected to DIMNM.
Overview of the equipment involved.
Researchers have demonstrated attacks on a number of services using chip manufacturers' TEE protection. (For ethical reasons, attacks were conducted against infrastructure that was identical to but separate from the targets' networks.) Some of the attacks included BuilderNet, dstack, and Secret Network.
BuilderNet is a network of Ethereum block creators that uses TDX to prevent parties from snooping on others' data, and to ensure fairness and fair redistribution of proof currency. The network creates blocks worth millions of dollars every month.
“We have demonstrated that an attacker with an attestation key can join BuilderNet and obtain configuration secrets, including the ability to decrypt confidential order flow and gain access to an Ethereum wallet to pay validators,” TEE.fail explains. “Additionally, a malicious operator could create arbitrary blocks or preempt (i.e. create a new transaction with higher fees to ensure their transaction is completed first) sensitive transactions for profit while maintaining deniability.”
To date, BuilderNet has not provided mitigation measures, according to the researchers. Attempts to contact BuilderNet representatives were unsuccessful.
Dstack is a tool for creating confidential applications that run on virtual machines protected by Nvidia Confidential Compute. The researchers used TEE.fail to forge evidence that the workload was running by TDX using Nvidia security. He also used “borrowed” evidence to falsify ownership of GPUs trusted by the relying party.








