Advancing security and ensuring privacy with machine learning
Source – helpnetsecurity.com
The Internet has many issues: lack of encryption and its governance, questionable marketing techniques, a misinformed average user. These issues are as old as the Internet itself. And machine learning algorithms can become the right tool to solve them.
1999 to 2018: Better connectivity, legacy technology
Back in 1999, only 4 percent of the world’s population was online. Today the number has reached 49%, and it just keeps growing. The migration from the offline world was followed by a rapid development of numerous online services and the advancement of connected devices.
The growth was so fast that most of the new industries didn’t implement adequate processes to ensure privacy and security.
This is not surprising knowing that getting hacked is rather common. Infections can be caught by merely opening the wrong email. Sharing information with a seemingly reputable company online could result in sensitive data leaks. Any device could become a zombie in a botnet without showing obvious signs of it.
Numerous data breaches threaten identities and bank accounts. Personal data is sold for anything from $50 for health records to $1000 for bank account information on the deep web. 91% of Americans agree that people have lost control over the collection and usage of personal information.
The technology has evolved, malicious actors are catching up swiftly, but the security is lagging behind.
Artificial intelligence as a tool to ensure privacy and security
Artificial Intelligence (AI) today creates a lot of heated discussions. It is seen as a marketing tool, an alternative term for statistical analysis, or an overhyped magical cure for all the tech problems. Yet, the key message is that after years of development, machine learning algorithms finally are delivering valuable results.
It is used in search engines, image classification, voice recognition, and many other areas. Its usage is growing in medicine, communications, transport, and gaming. Even in its early stages, AI is delivering results that are hard to compare with insights human analysts come up working without the help from machines.
The discussion is shifting: the internet is moving away from using AI as a buzzword towards embracing it as a tool.
How can AI algorithms advance user privacy and security?
The key issue that cybersecurity companies face today? It’s striking a balance between security and privacy. DNS blacklisting is the most popular legacy method to ensure this. It involves having a database of malicious websites and blocking those websites on user computers.
To achieve that, the legacy security companies have to read the communication between the user computer and the web server where the website is stored. The process is called Deep Packet Inspection (DPI). During it, user privacy might not be preserved: it’s the equivalent of your postman reading your letters.
The metadata that is unencrypted contains enough information to provide the necessary protection to the end user. Security can be provided without DPI by analyzing only the metadata: a small portion of data that can be found in the header (the “label”) of the data packet. Security companies can use machine learning, create behavioral profiles, and use only the metadata of the packets that travel from the user to the website servers.
Why artificial intelligence?
Security algorithms powered by AI are beneficial because of three reasons:
- Speed. Training such algorithms is much faster than collecting rules that define malicious actors.
- Precision. Copious amounts of data can be analyzed easily; therefore more precise results are delivered. It also enables personalization.
- Privacy. You don’t need to use large amounts of data, or intrusive methods such as DPI, to provide the results.
The key takeaway here is that Artificial Intelligence is a tool. It has its limitations and even at the current speed of advancement, AI is not going to replace highly skilled human workers anytime soon.
That said, artificial intelligence can help businesses to provide better service for their users while keeping their privacy intact. It can help create that balance that legacy strategies are missing.
About CUJO AI
CUJO AI is the leading artificial intelligence company providing network operators AI-driven solutions, including AI security, advanced device identification, advanced parental controls, and network analytics. CUJO AI Platform creates intuitive end-user facing applications for LAN and wireless (mobile and public wifi). Each solution can be implemented as a white-label offering. CUJO AI was recently listed as a “Vendor to Watch” and a “Cool Vendor in IoT security” by research company Gartner. In May 2018, the company has closed a strategic Series B round, led by Charter Communications, valuing the company in access of $100M. CUJO AI was selected as one of the World Economic Forum’s Technology Pioneers 2018.