As web-based attacks increase, the ability of conventional security software to keep up with them is limited. Malware delivery, phishing, and other attacks such as cross-site scripting all occur via the web, and the pages and URLs that deliver them are nearly infinitely flexible. This all combines to make web-based threats difficult or impossible for conventional security-based software to catch early, often relying on detecting the last stage of malware delivery or account hijacking after a phishing page has led to stolen credentials.
Sophos AI is working to make the web a safer place, using deep neural networks to detect malicious URLs, detect and warn users about phishing sites, and block malware delivery at the source. These deep learning models add additional layers of security to Sophos’s synchronized security architecture, stopping threats before they ever reach your network.
Check out our peer reviewed publications below:
URL security analysis: https://arxiv.org/abs/1702.08568
HTML security analysis: https://arxiv.org/abs/1804.05020
We’re also developing next-gen content classification models that combine image recognition, URL analysis, and web page content analysis to classify websites according to their content. Check out the demo here [DEMO]