Cloud Sandbox
Protect your users from previously unknown malware. A neural network over 8,600 file attributes, continuously trained, seamlessly built into web and email scanning.
Catch the malware no signature has seen.
Consisting of over 8,600 file attributes, our Cloud Sandbox includes an intelligent neural network to rapidly determine the presence of new, unknown malware. It leverages machine learning to continuously improve with every new file uploaded for analysis, so the radar sharpens itself instead of waiting for the next signature update.
Benefits.
Defend against the unknown
An additional layer of web and email security protects you against brand new or unknown, sophisticated threats.
Leading technology
Analyses over 8,600 file attributes and is continuously trained for an effective, up-to-date malware radar.
Built-in protection
Seamlessly integrated into standard file scanning of the Secure Web Gateway and Secure Email Gateway.
Trained on malware, retrained every day.
A neural network decides whether a file contains malware. More than 8,600 file attributes feed it, and a large, labelled data set of malicious and clean files teaches it to tell one from the other.
- Every unknown file is judged against everything the network has seen before
- With each upload the network gets smarter as machine learning tunes its evaluation to the newest hallmarks of malware
- Re-learning mechanisms keep the neural network accurate over time
When Cloud Sandbox comes into play
Beyond conventional malware scanning, the Cloud Sandbox runs a fast, layered decision so most files never need a full analysis:
1 · Hash Cache
A unique file hash is calculated for the attachment or download and checked against a local cache of known-good and known-bad hashes.
2 · Hash Database
If the local cache does not know the file, the hash is looked up in the large, global cloud database.
3 · Risk Score
If the hash is nowhere on record, a local engine scores the file with machine-learning heuristics. Only files above the threshold are uploaded to the cloud.
4 · Machine Learning
If heuristics are not decisive, the file is evaluated against more than 8,600 attributes and a tailored ML module determines whether it is malicious.
Go deeper.
Leave complexity behind.
See how the Open Systems SASE Experience protects your users against the malware no one has catalogued yet.