Blog Zscaler
Ricevi gli ultimi aggiornamenti dal blog di Zscaler nella tua casella di posta
IscrivitiAI/ML-Powered URL Classification: Categorize and Conquer
If you ever try wrapping your head around the sheer enormity of the internet, your brain might start to hurt. The indexed internet teems with roughly 4.5 billion unique web pages at any moment, with countless subpages. That's almost 4.3 billion IPv4 and more than 340 undecillion (that’s 36 zeroes) IPv6 addresses!
Add to that the explosive growth of generative AI capabilities. Like the Usain Bolt of website creation, GenAI is sprinting ahead to build sites faster than you can say “HTML.”
But all these lightning-fast developments have left URL block lists obsolete and many organizations vulnerable to evolving web threats.
With a gazillion domains, URLs, sites, and IP addresses, getting lost in the internet's forbidden corners is almost inevitable. That’s precisely why understanding URLs is crucial for IT security.
Key component of your GRC Framework
URL categorization ties directly into the ongoing stampede of businesses trying to align their security posture with the Governance, Risk, and Compliance (GRC) framework.
Governance
Part of effective governance entails managing employee productivity. According to a recent study, if done right, each 1% increase in employee productivity can lead to a 1% increase in shareholder value.
URL categorization helps businesses achieve this by restricting access to non-work-related websites, fostering a more focused work environment.
Moreover, with clearly defined web governance policies, businesses don’t have to trust that employees will know how to navigate the web safely; instead, they can guarantee it!
Risk
In addition to identifying and blocking access to sites known for hosting malicious content, phishing scams, or malware, and helping the security team prevent botnet callbacks, cross-site scripting, and so on, URL categorization can support incident response efforts by providing valuable information about the source of an incident.
By analyzing URL access logs, organizations can trace the origin of malicious activities, identify affected systems, and take appropriate remediation measures to contain an incident and prevent further damage.
Compliance
URL classification enables businesses to categorize websites associated with embargoed countries or containing content restricted by trade embargoes as high-risk or prohibited content.
For example, financial institutions may use URL classification to block access to gambling websites to comply with regulations such as the PCI DSS.
Similarly, it can assist in identifying and blocking access to websites hosted in jurisdictions with inadequate data protection laws or surveillance practices, in line with the Schrems II regulation.
Lastly, URL classification and categorization systems can generate reports on web usage patterns, blocked sites, and compliance violations. This provides valuable data for compliance reporting purposes—a must-have capability required by many regulatory bodies.
Where traditional approaches fall short
Although legacy approaches to URL categorization may have been effective in the past, they often fall short in addressing the complexities and scale of the modern web.
1. Lack of the secret (AI/ML) sauce
Manual URL categorization is nearly impossible in the new AI/ML-driven world order.
A manual approach will quickly run into speed, cost, and scale constraints. Plus, different people may categorize URLs differently based on their personal biases, experiences, and interpretations, leading to inconsistencies and inaccuracies.
2. Size doesn’t matter
Once upon a time, IT teams boasted extensive lists of IPs and sources as an effective way to track and block network connections to malicious sources—the bigger the list, the better the flex.
Not anymore.
Firstly, these static lists are no match for the dynamic nature of the internet. Relying on IP block lists for categorization is bound to generate false positives and negatives.
Secondly, many websites are hosted on shared hosting platforms or content delivery networks, meaning multiple websites may share the same IP address. Categorizing URLs solely based on IP addresses can lead to inaccuracies and misclassifications.
3. Boiling the ocean
What if we crawl every web page, file, and link there is for malicious content? After all, ChatGPT does something similar to gather information.
Maybe. However, crawling websites without permission may violate terms of service or copyright laws. Some websites explicitly prohibit web crawling through their robots.txt file or terms of service agreements.
Web crawlers are ideally used in conjunction with other approaches, such as machine learning, keyword analysis, reputation systems, and human moderation, to achieve more accurate and comprehensive URL categorization.
Zscaler AI/ML-Powered URL categorization
Zscaler offers powerful URL categorization as a native feature of Zscaler Internet Access™, the world’s most deployed security service edge (SSE) platform, which includes industry-leading secure web gateway, data loss prevention, cloud-gen firewall, and more.
Our powerful URL categorization leverages AI and machine learning to analyze uncategorised URLs in real time. Based on dynamic content analysis, ZIA then assigns appropriate categories, making sure policies are enforced consistently.
This ensures that the majority of the URLs accessed via our cloud are categorized—in addition to expanding our URL categorization coverage and accuracy.
How does URL categorization work?
- The user forwards internet-bound network traffic to the closest available Zscaler Service Edge (SE).
- The SE downloads the user policy from the Central Authority (CA) for inspection and processing of the URL.
- The next step is deciding whether the network traffic needs to be decrypted for SSL inspection inside the Zscaler cloud or not.
- If SSL inspection is disabled or not applied to that location, the HTTP request will be forwarded for URL filtering inspection.
- The Zscaler service intercepts the HTTPS request, and through a separate SSL tunnel, sends its own HTTPS request to the destination server and conducts SSL negotiations with the user’s browser, validating the certificate chain.
- Inside the Zscaler cloud, the URL filtering engine filters content by criteria, and the URL policies are evaluated and applied based on configuration.
Predefined classes and super-categories
Zscaler provides a comprehensive hierarchy of categories, including Business & Economy, AI and ML applications, Social Networking, Entertainment, and many more. This lets you create granular policies, blocking, allowing, or monitoring access to specific categories.
In addition to predefined categories, you can create custom categories based on URLs or IP addresses, keywords, and IP ranges. With URLs or IP addresses, you can block specific websites. With IP ranges, you can block a specific range of IP addresses for websites.
With keywords, you can block websites based on any words that may appear in a URL. For example, you could block all websites with the term “gambling” in the URL.
Zscaler URL categorization also enables you to create policies based on top-level domains (TLDs).
There’s more.
In addition to all the features and capabilities just discussed, Zscaler AI/ML URL categorization also facilitates:
- Browser isolation: Zscaler Browser Isolation works in tandem with URL filtering to render websites in an isolated environment—protecting user devices from potential malware or exploits. Users can still interact with the content, but any malicious code is contained within the isolated session—also isolating uncategorized or unknown sites.
- Identification of newly created and revived domains: Zscaler can identify and categorize newly registered and revived domains that usually pose a higher security risk—preventing them from passing through traditional filtering methods.
- Safe search integration: Zscaler works with popular search engines to enforce safe search options. This helps users avoid encountering inappropriate content during web searches.
- Compliance with the Children's Internet Protection Act (CIPA): Zscaler classification aligns with CIPA regulations, ensuring a secure browsing experience for educational institutions and organizations that work with children.
- Embedded sites categorization: Our URL categorization engine can also identify sites that have been translated using translation service websites.
Is your business planning to invest in URL categorization?
Our experts can help you understand the options that best fit your unique requirements. Ready to start a conversation? www.zscaler.com/company/contact
To see our URL categorization in action, watch this demo.
Questo post è stato utile?
![dots pattern](/_next/image?url=%2Fassets%2Fimages%2Fcta_bgpattern.png&w=1920&q=75)
Ricevi gli ultimi aggiornamenti dal blog di Zscaler nella tua casella di posta
![](/_next/image?url=%2Fassets%2Fimages%2Floading.webp&w=256&q=75)
Inviando il modulo, si accetta la nostra Informativa sulla privacy.