Cutting through the hype: AI and ML for the security space

Today’s organisations face numerous diverse threats to their people, places and property, sometimes simultaneously. Security leaders now know all too well how a pandemic can cripple a company’s ability to produce goods and services, or force production facilities to shut down, disrupting business continuity. For example, a category three hurricane barreling towards the Gulf of Mexico could disable the supplier’s facilities, disrupt the supply chain and put unexpected pressure on an unprepared local power grid.

Delivering timely critical information

Tracking such risk is hard enough, but managing it is even more difficult. A swift response depends on delivering the right information to the right people, at the right time. And, it’s not as easy as it sounds. Indeed, 61 percent of large enterprises say critical information came too late for them, in order to mitigate the impact of a crisis, according to Aberdeen Research (Aberdeen Strategy & Research).

These challenges are accelerating the hype around Artificial Intelligence (AI)

These challenges are accelerating the hype around Artificial Intelligence (AI). The technology promises to help us discover new insights, predict the future and take over tasks that are now handled by humans. Maybe even cure cancer.

Accelerating the hype around AI

But is AI really living up to all this hype? Can it really help security professionals mitigate risk? After all, there’s a serious need for technology to provide fast answers to even faster-moving issues, given the proliferation of data and the speed at which chaos can impact operations.

Risk managers face three major obstacles to ensuring business continuity and minimising disruptions. These include:

• Data fatigue - Simply put, there’s too much data for human analysts to process in a timely manner. By 2025, the infosphere is expected to produce millions of words per day. At that pace, you’d need an army of analysts to monitor, summarise and correlate the information to your impacted locations, before you can communicate instructions. It’s a herculean task, made even more difficult, when we consider that 30 percent of this global datasphere is expected to be consumed in real time, according to IDC.
• Relevance and impact - Monitoring the flood of information is simply the first hurdle. Understanding its impact is the second. When a heat dome is predicted to cover the entire U.S. Pacific Northwest, risk managers must understand the specifics. Will it be more or less hot near their facilities? Do they know what steps local utilities are taking to protect the power grid? Such questions can’t be answered by a single system.
• Communication - Once you know which facilities are impacted and what actions to take, you need to let your employees know. If the event is urgent, an active shooter or an earthquake, do you have a fast, effective way to reach these employees? It’s not as simple as broadcasting a company-wide alert. The real question is, do you have the ability to pinpoint the location of your employees and not just those working on various floor in the office, but also those who are working from home?

How AI and ML cut through the noise

Although Artificial Intelligence can help us automate simple tasks, such as alert us to breaking news, it requires several Machine Learning systems to deliver actionable risk intelligence. Machine Learning is a branch of AI that uses algorithms to find hidden insights in data, without being programmed where to look or what to conclude. More than 90 percent of risk intelligence problems use supervised learning, a Machine Learning approach defined by its use of labelled datasets.

The benefit of supervised learning is that it layers several pre-vetted datasets, in order to deliver context-driven AI

The benefit of supervised learning is that it layers several pre-vetted datasets, in order to deliver context-driven AI. Reading the sources, it can determine the category, time and location, and cluster this information into a single event. As a result, it can correlate verified events to the location of the people and assets, and notify in real time. It’s faster, more customised and more accurate than simple Artificial Intelligence, based on a single source of data.

Real-world actionable risk intelligence

How does this work in the real world? One telecommunications company uses AI and ML to protect a mobile workforce, dispersed across several regions.

An AI-powered risk intelligence solution provides their decision makers with real-time visibility into the security of facilities, logistics and personnel movements. Machine Learning filters out the noise of irrelevant critical event data, allowing their security teams to focus only on information specific to a defined area of interest. As a result, they’re able to make informed, proactive decisions and rapidly alert employees who are on the move.

Four must-have AI capabilities

To gain real actionable risk intelligence, an AI solution should support four key capabilities:

• A focus on sourcing quality over quantity. There are tens of thousands of sources that provide information about emerging threats - news coverage, weather services, social media, FBI intelligence and so much more. Select feeds that are trusted, relevant and pertinent to your operations.
• Swift delivery of relevant intelligence. To reduce the mean-time-to-recovery (MTTR), risk managers need an accurate understanding of what’s happening. Consider the different contextual meanings of the phrases ‘a flood of people in the park’ and ‘the park is at risk due to a flood’. Machine Learning continuously increases the speed of data analysis and improves interpretation.
• Ability to cross-reference external events with internal data. As it scans different data sources, an AI engine can help you fine-tune your understanding of what’s happening and where. It will pick up contextual clues and map them to your facilities automatically, so you know immediately what your response should be.
• Ready-to-go communications. Long before a threat emerges, you can create and store distribution, and message templates, as well as test your critical communications system. Handling these tasks well in advance means you can launch an alert at a moment’s notice.

The ability to minimise disruptions and ensure business continuity depends on speed, relevance and usability. AI and ML aren’t simply hype. Instead, they’re vital tools that make it possible for security professionals to cut through the noise faster and protect their people, places and property.