In an exclusive interview with Sridhar Muppidi, IBM Fellow and CTO, IBM Security at Cybertech Israel 2019, shares his view on IBM’s implementation of AI in security solutions and it’s road map on cybersecurity.
1) Could you share the IBM’s approach on implementation of AI in cybersecurity?
AI is used in cybersecurity to primarily detect threats, understand risk, followed by consolidation of intelligence and definitely for orchestration of information. We at IBM are using various machine learning processes to detect threats faster and better. We do not use AI for fancy things, but to reduce number of false positive for application testing and to use AI for behavioural biometrics. And as we go then into the next section, which is consolidation for intelligence, we already started with Watson, which is way to teach the language of security to the computer. We use natural language processing as a way to collect the entire security information from the world and put it in a way that we can stat learning or questioning. IBM is into security business for a long time and so we understand the best way protecting passwords and sensitive information and hence we use that expertise to teach computers using AI and machine learning as a way to be trusted advisors. If you find any person risky and you are desperate to protect information from that person or any untrusted party or resources, then these advanced computers with the help of AI can help you to do so. We are using AI as adversarial AI, which is basically utilization of AI to defeat AI. We are using AI for launching an attack, using AI to steal AI. So we are using AI to retrain existing AI,. You can easily fool a system, which is doing facial recognition with some blind spots and fool the system to be somebody else. You can steal the data that is used to train AI and falsify information. So, we have to be more matured in that area and IBM is already with that momentum. We have recently published an open-source toolkit, called adversarial robustness toolkit, so that we can collaborate faster with universities and industries.
2) To what extent the collaboration works? Can Institutions open up to the expectations?
We can not do everything by our own. To invent a better and effective solution, we have to collaborate. Need of the hour is intelligent way of collaboration and information orchestration. We at IBM, open up the interfaces to make sure that products are talking to each other. For sure, we don’t want to open up the deep secrets but doing it in a manner to safeguard the data privacy, making sure to share the insights but not the way we do it. So, you have to safeguard the intellectual capital, data privacy loss, but at the same time we have to bring partners together, so that we create an outcome.
3) How does IBM predict in terms of AI in cybersecurity?
We at IBM are very matured in the use of AI in cybersecurity. We have Watson for cybersecurity analysis, which is a huge graph of all the information. So, based on that we start predicting things ahead of real damage. If anything is happening in China, India and Sao Paulo, we are able to pull out information together and correlate and stop from propagating. We are going forward towards predictability with certain level of confidence and applicability. For example, if any new malware is found, we try to predict the damaging capability of the malware. We at IBM are trying to answer the unpredictables and predicting the future. Beyond that, we are using IoT to be able to apply the same expertise and we are also focusing significantly on quantum computing as well as deception techniques, which is the next generation of technology.