IBM Watson’s Fight Against Cybercrime

Analytics & Cognitive - Other
  • Smaller Small Medium Big Bigger
  • Default Helvetica Segoe Georgia Times



Watson for Cyber Security provides new tools for overwhelmed security staff.

In any IT system, how do you know if it’s under attack?

That’s really a loaded question. Systems are always under attack. Day and night, whether you’re aware of it or not, people, machines, and programs are hitting any device that can be reached either on the public Internet or on private networks. Old NAT rules on firewalls can be very effective holes to constantly probe a commonly used port on internal servers. Internal employees with computers infected with many forms of malware are hitting every machine on corporate networks in countless businesses all over the world right now.

The plight of the IT administrator or security administrator is a complete and utter data overload. In companies with modest budgets, the ability to identify security attacks is quite limited.

Data theft is one of the hardest things to uncover. If an attacker downloads a file, he or she will leave the old file where it was and attempt to cover their tracks. They won’t delete the old file. That’s a dead giveaway that someone was there.

When does IT usually start to investigate a security problem? When it’s identified, or at least when the symptoms are evident enough to trigger an action. That usually means that an application, server, or service is running far slower than usual or is not running at all. Think of a distributed brute force attack against an SMTP server port. Depending on the attack load strength, the server will likely slow down. If enabled, log files will show authentication failures coming from the same IP addresses against accounts that largely don’t exist. The usual symptom is usually identified by the users as the mail system is slow. In IT, that’s a common complaint and doesn’t trigger fears for the worst. Especially mail. Unless you’re viewing mail online in a browser in an interactive fashion, you’re not going to see the server behave slowly in real time. If you’re using a product like Outlook or Notes, the send/receive is done in the background and users won’t see much difference. Hours later, users are complaining that they haven’t received an email they know someone sent, and that can end up being the real indicator of a problem. In this scenario, IT is far behind the eight ball. In a matter of hours, it’s very possible that an account with weak credentials has already been compromised and the server is being hijacked to push spam to the world. Or worse.

While SMBs are the target du jour for attackers due to their lack of staff and expertise, larger enterprises with dedicated security staff are likely in no better shape. The more infrastructure you have, the more you need to monitor, manage, and defend. The bigger you are, the more likely you have a target on your back. There’s an old saying about when you get big enough to look over your fence is the point when people start throwing rocks at you.

In February, IBM announced Watson for Cyber Security. Watson has been trained in the past year to understand the language of cyber security, learning the contents of over one million security documents. According to IBM, the average security team has to manage over 200,000 security events per day and is wasting approximately 20,000 hours per year on false positives. These numbers are obviously describing larger enterprises than an average five-person IT department.

Watson for Cyber Security will be integrated into IBM’s new Cognitive SOC product, bridging security operations with cognitive computing. This will allow customers to manage attacks across devices, networks, and even the cloud. Watson for Cyber Security is designed to find connections between obscure data points and allow companies to quickly and intelligently react to valid attacks. Since it’s a learning solution, it will get smarter over time and be able to more accurately and proactively combat threats. A component of Cognitive SOC is called QRadar Advisor for Watson, which combines IBM’s security analytics platform QRadar with Watson in order to investigate and qualify security events or potential security events. Another component of Cognitive SOC is called IBM BigFix Detect, which is an endpoint detection and response (EDR) solution. BigFix Detect extends cognitive processing to network endpoints where it targets malicious behavior and attempts to mitigate it before it spreads to the rest of the network.

"The Cognitive SOC is now a reality for clients looking to find an advantage against the growing legions of cybercriminals and next-generation threats," said Denis Kennelly, Vice President of Development and Technology, IBM Security. "Our investments in Watson for Cyber Security have given birth to several innovations in just under a year. Combining the unique abilities of man and machine intelligence will be critical to the next stage in the fight against advanced cybercrime."

QRadar Advisor for Watson is already in use at Avnet, University of New Brunswick, and about 40 other customers worldwide.

Sean Valcamp, Chief Information Security Officer at Avnet, said, "Today's sophisticated cyber security threats attack on multiple fronts to conceal their activities, and our security analysts face the difficult task of pinpointing these attacks amongst a massive sea of security-related data. Watson makes concealment efforts more difficult by quickly analyzing multiple streams of data and comparing them with the latest security attack intelligence to provide a more complete picture of the threat. Watson also generates reports on these threats in a matter of minutes, which greatly speeds the time between detecting a potential event and my security team's ability to respond accordingly."

IBM has also invested in tools for its X-Force Command Center, including a Watson-powered chatbot for IBM Managed Security Services customers. As well, IBM announced a new research project, code named Havyn, which is a voice-powered security assistant powered by Watson’s natural language technology to interact with security professionals.

Project Havyn uses Watson APIs and BlueMix to provide real-time response to verbal requests and commands using data from customer-specific security data and open-source security intelligence, including IBM X-Force Exchange. Havyn can answer questions about recent security threats and provide remediation tasks. Imagine being able to ask a computer, in natural language, “How can I alter my company’s environment to fight recent threats that have appeared this week?” That is a game changer.

In fact, I would suggest that Watson for Cyber Security can potentially be the most significant breakthrough in combating cyber threats in the last 10 years. Society pivots on major breakthroughs in science, whether it’s the advent of vaccinations, the steam engine, or air travel. Combating cyber crime has, until now, been largely a reactive process worldwide. We may be at a point where effective and automated proactive monitoring and remediation can occur, allowing IT to focus on what it does best: adding business value.