AWS introduces Mithra. This is not a paid service, but a huge neural network that the cloud provider uses internally to quickly detect malicious domains.

About a quarter of all Internet traffic runs through Amazon Web Services systems every day. This immense scale means that the cloud provider itself has to design and develop systems to protect itself, its customers and the Internet. Against this background, AWS is introducing Mithra today.

Mithra is not a service you can purchase, but an internal system from AWS itself. The system consists of a self-developed neural network that uses algorithms for threat intelligence. It has 3.5 billion node and 48 billion edgeand can quickly check domains for legitimacy.

Huge data volume

In a single region, AWS processes up to 200 trillion requests per day. Mithra can detect an average of 182,000 new malicious domains. These are given a rating and placed on a list that AWS and other parties can use.

The cloud provider states that Mithra AI can detect threats days, streets, or even months earlier than other systems from third-party security providers. For example, GuardDuty is based on the qualitative list of unwanted domains. Based on Mithra’s data, GuardDuty can block domains and thus protect customers.

Other services that use external systems can include Mithra’s list. According to AWS, this would help reduce false positives. Finally, people also benefit from Mithra’s analytics and scores. They can use the data for their research.

Only with AI

According to AWS, AI is critical to the type of protection Mithra provides. AI looks at domains and indicators of their reliability. This data is in a graph database so large that a human cannot understand it, let alone analyze it. AI is great at identifying patterns and irregularities in this data.