Macie Brings AI-First Infrastructure to AWS

Artificial intelligence(AI) is becoming an integral component of all aspects of enterprise architectures and infrastructure is not the exception. At the recent AWS Summit in New York City, Amazon announced the release of Macie, a new cloud service that leverages machine learning to protect data in the AWS cloud. Macie is an example of a new trend that I often referred to as AI-first infrastructure.

Conceptually, Macie is designed to identify and protect sensitive information across a broad range of AWS data services like S3. When users configure Macie, they use a risk score to classify sensitive information. The initial classification effectively turns into a training dataset used to help Macie get familiar with the structure of the data and apply the knowledge to new data going forward. From that point on is where the magic of Macie happens. The platform leverages unsupervised learning algorithms to detect access patterns in sensitive data elements. Anomalies in the access behavior are flagged and communicated to human experts for subsequent reviews.

In This Really New?

The concept behind Macie contains some interesting novelties but is far from being a breakthrough innovation in the cloud security space. The idea of leveraging machine learning models top detect access control patterns in cloud data systems has been around for a few years. Recently acquired Elastica has been one of the pioneers in the space that today include incumbents such as Microsoft which has acquired similar capabilities and implement them into its Azure and Office365 platforms. However, there are some specific benefits of Macie brings to the table that should not be taken lightly.

First of all, Macie will be able to provide access control intelligence across the large portfolio of data storage services available in the AWS cloud. At the moment, Macie only works with S3 but new storage services should soon be supported. Similarly, it is not crazy to think that, in the future, Macie could be integrated with third party cloud database or services running on AWS. Secondly, I believe Macie’s laser focus on access control patterns can drive a lot of innovation and benefits to AWS storage systems while avoiding the distraction of tackling broader cyber-security areas. Finally, we should ignore the fact that AWS’s market dominance and large footprint within enterprise customers can help to accelerate the adoption of Macie to make it a leader in the space.

The Raise Towards and AI-First Infrastructure

In the past, I’ve written extensibly about the concept of AI-First infrastructure technologies. Macie is not the first example of AI-First infrastructure stack but is, arguably, one of the most visible attempts to take those concepts mainstream.

By AI-First infrastructure, we refer to technologies that incorporate many of the knowledge of devops and IT operators into the infrastructure layer. Security, storage networking, compute are some of the infrastructure areas that are being re-imagined with AI as a foundational building block. Obviously, the AI-infrastructure space offers plenty of opportunities for startups but they should be aware that cloud incumbents such as Amazon, IBM, Microsoft and Google are jumping aggressively into that market.