AMD Infinity Guard
The prevalence of ransomware attacks and data breaches in recent years has made it difficult for important business sectors to collaborate. Reports state that organizations are unable to work with suppliers who are attempting to develop potentially ground-breaking apps or discoveries due to the risk posed by threat actors. To keep up the strict restrictions necessary for specific data sets, some businesses don’t even share data within. Researchers’ failure to obtain vital data impedes their capacity to conduct significant study in a number of fields, including government, banking, and healthcare.
Healthcare AI acceleration via a safe platform for algorithm creators and data custodians to collaborate
Data is Never Exchanged or Viewed
The data steward’s safe, HIPAA-compliant environment is where the data is never removed.
Processing Real-World and Protected Data
Employs primary data, which comes directly from the source, as opposed to artificial or de-identified data. Every time, the data is encrypted.
Never Is Intellectual Property Seen or Shared
The algorithm is always encrypted, both when it is uploaded to EscrowAI and when it is moving through the container to the data steward and inside the protected environment of the data steward.
Technology with Secure Enclaves
EscrowAI uses secure enclave technology to reduce the possibility of algorithm IP questioning and data exfiltration during computing.
Matchmaker and intermediary
BeeKeeperAI reduces the time, effort, and expenses of data projects by more than 50% by serving as a matchmaker and broker between data stewards and algorithm developers.
Alan Czeszynski, an expert in the security industry and the marketing and product development leader at BeeKeeperAI, was gracious enough to join me on the AMD EPYC TechTalk podcast series following the Confidential Computing Summit industry gathering in San Francisco. They talked about the state of security and how there has never been a greater need for better hardware and software safeguards.
BeeKeeperAI
EscrowAI, a technology that combines private and confidential computing technology to allow software developers, data scientists, and data owners to collaborate in trusted execution environments (TEE), is utilized by San Francisco-based BeeKeeperAI.
The technology of BeeKeeperAI ensures that an owner always has control over their data. In addition to offering end-to-end encryption and algorithmic and model encryption to safeguard intellectual property, BeeKeeper also applies the algorithm to the data. The business establishes a TEE in a cloud data storage environment after an algorithm is prepared to run against data. Consequently, the data is cut off from all stakeholders, including BeeKeeperAI, the cloud service provider, the data owner, and the owner of the algorithm.
Nobody can see what goes on within the TEE; everyone can only access the output to which they are legally permitted.
“Bring these parties together to enable development and testing of artificial intelligence and machine learning models,” according to Alan, is made possible by BeeKeeper’s secure environment.
Big large language models (LLMs) and generative AI have gained popularity, and as a result, businesses are now more conscious of the need to secure AI, according to Alan. Protecting every stage of the AI and machine learning lifecycle has received a lot of attention lately. According to Alan, this is one of the reasons private computing is starting to get a lot of traction.
Alan warns that legacy security solutions might not provide enough protection in the AI era. The problem with LLMs is that they essentially turn into enormous repositories of all your secrets if you wish to locally train them on your own data,” he continued.
While CISOs and IT administrators prioritize data protection, business managers and data scientists frequently place greater importance on obtaining the data required to develop models that improve the company. Alan claimed that it is far too common for the procedure of obtaining private, protected data to be difficult, costly, and time-consuming. He described a few of the intricate details.
It is usually necessary for parties to have detailed, extremely formal data-use agreements in place. There are often several restrictions on how the data can be interacted with. Audits have to be done, and they always have to. BeeKeeperAI eliminates the effort by offering a technical answer to many of these security challenges.
“Their goal is to eliminate that from the end user and basically take it upon selves,” Alan stated. “The platform then allows the true value, which is basically secure collaboration, getting access to the data, developing your models, being able to execute your AI, ML lifecycle in a secure environment.”
Alan acknowledged that the security features incorporated by AMD EPYC CPUs had strengthened BeeKeeperAI’s offerings. AMD Infinity Guard includes these technologies, such as Secure Encrypted Virtualization and Secure Nested Paging, or SEV-SNP. They prevent the contents of a virtual machine’s memory from being accessed by other VMs operating on the same system or the server they are operating on.
Alan also mentioned adaptability, which is another significant advantage of AMD EPYC. AMD have to provide [clients] a variety of possible platforms, and EPYC is a fantastic one,” said Alan. “In those situations, the safe paging feature of encrypted virtualization and secret containers or virtual machines based on the EPYC CPU is quite advantageous. One of the main advantages of utilizing EPYC processors is that algorithm developers no longer have to adhere to any certain OS type thanks to this lift-and-shift technique.”
