Intel and AMD trusted enclaves, the backbone of network security, fall to physical attacks

In today's cloud computing landscape, the integrity of data protection mechanisms embedded in processors from Intel and AMD is crucial for safeguarding sensitive information from potential cyber threats. These mechanisms, known as Trusted Execution Enclaves (TEEs), are designed to keep confidential processes secure from unauthorized access, particularly within data centers. Prominent cloud service providers advocate for the use of these protections, which include Intel's Software Guard Extensions (SGX) and AMD's Secure Encrypted Virtualization with Secure Nested Paging (SEV-SNP). However, recent studies have highlighted significant vulnerabilities in these security features that put user data at risk. On Tuesday, two independent research teams unveiled their findings, detailing novel attack methods that undermine both SGX and SEV-SNP. One of the attack strategies, referred to as Battering RAM, not only enables attackers to view encrypted data but also allows them to alter it, potentially introducing backdoors or corrupting critical information. Another attack, named Wiretap, demonstrates a passive approach, successfully decrypting sensitive data secured by SGX without detection. Both attacks leverage a specialized piece of hardware called an interposer, which is positioned between the CPU and the memory module, allowing it to monitor data transfers. These exploits take advantage of the deterministic encryption methods employed by both Intel and AMD, which result in the same ciphertext being produced from identical plaintext under the same encryption key. This predictability in encryption compromises the overall security framework intended to protect user data in cloud environments.