Artificial intelligence a real game-changer for the security space
Just a few years ago, artificial intelligence was the domain of science fiction. But today it is a reality that some experts are predicting will have a greater impact on our lives than the industrial revolution.
No industry will be spared the disruptive effects of AI, and forward-thinking businesses are already preparing for it. A recent survey by IT research firm Infosys found over 85 per cent of large organisations worldwide already have middle or late-stage AI developments.
The security sector is no exception, and companies in both the physical and cyber security spaces are already leveraging AI in order to gain a competitive advantage.
Where cyber security comes in
Global demand for cyber security expertise is exploding, and by some estimates , there will be a shortage of two million cyber security experts worldwide by 2019. But AI promises to help fill the gap.
By trolling through vast amounts of data and then parsing it in a fraction of the time humans would need to do the same job, AI-driven software is revolutionising cyber security. It is moving the focus away from strengthening perimeters and locking out threats, to instead detecting bad behaviour the instant it begins.
For example, the cyber AI of Darktrace , created by mathematicians from the University of Cambridge, runs on algorithms that mimic the human immune system. They learn what “normal” looks like for any network, be it physical, virtualised or cloud, and then report any deviations and anomalies in real time.
In the realm of physical security, AI is giving security systems digital brains to match their digital eyes.
For example, the video analytics solutions of Australian tech firm Brainchip aid law enforcement and intelligence organisations by rapidly searching vast amounts of video footage to identify faces or other objects. With the ability to process 16 channels of video simultaneously, or 600 images a second at over 99 percent accuracy it far exceeds the limits of human ability.
The Company has developed a technology called Neuromorphic computing that learns, evolves and associates information just like the Human Brain.
Neuromorphic computing is a branch of artificial intelligence (AI) that simulates the functionality of the human neuron.
BrainChip’s revolutionary spiking neural network (SNN) technology has many attractive characteristics, including the ability to be trained instantaneously (“one-shot learning”), high accuracy and low compute overhead.
This is an important feature in the world outside of the internet, where massive datasets are not available. For instance, a police department looking for a suspect in live video streams does not have thousands of images of that suspect, nor does it have weeks to train a traditional convolutional neural network system.
BrainChip’s civil surveillance solutions provide law enforcement, homeland security, and intelligence agencies with the ability to rapidly identify objects or faces in large amounts of archived or live streaming video.
BrainChip’s technology powers a custom card and chip recognition system that has been successfully trialled and deployed in major casinos where cards can be dealt naturally, without having to conform to any specific orientation. The card is recognized as long as the card is placed on the table with the suit facing upwards, as is the case in a Baccarat game.
AI makes it possible for computers to assess situations and events and make decisions based on information from multiple, unrelated sources. For example, video analytics could be used in conjunction with voice and facial recognition, security bulletins, data from PIR sensors and card readers to control building access.
Flare, an AI security system from German start-up BuddyGuard , uses biometrics and other inputs including a microphone and a database of dangerous sounds to decide whether a security issue warrants attention, and it can then summon emergency services autonomously.
Additionally, the Canary home security device monitors changes in air temperature and humidity, potentially working in conjunction with motion detectors to decide whether a window has been broken.
Glimpse of the future
The security system at the headquarters of Shanghai-based Yitu Technology provides a glimpse of the future . It uses AI, facial recognition software and multiple cameras to track everyone entering the building for the entire time they are inside.
But the technology that Yitu Technology sells to Chinese authorities is even more impressive. Dubbed Dragonfly Eye , it is built on a constantly expanding database of faces that already exceeds 1.8 billion, with over 300 million of those collected at border crossings such as airports.
On its first day of operation on the Shanghai Metro last year, Dragonfly Eye identified a wanted man as he entered a station and sent his photo to police, who then made an arrest.
In the following three months, a further 567 suspects were arrested on the city’s underground network.
Yitu’s chief executive Zhu Long warns those who are still debating whether AI is hype or reality are wasting their time—“it is already here’’. “AI is a revolution that will eventually surpass the scale and pace of the industrial one,” says Zhu.
Artificial intelligence, real concerns
The potential of AI-powered security systems is so great, it is hard to predict what abilities they may develop over time. They are constantly learning and improving their own performance. And they never stop working—AI never gets sick or needs to eat, sleep or take leave.
But the advent of these increasingly intelligent and intrusive systems also raises serious questions around social justice, not to mention some potentially thorny legal issues where computers are making critical decisions on behalf of physical businesses.
Privacy will be a major concern too. The manufacturers of some AI systems are already addressing this: for example, BuddyGuard’s Flare home security monitor uses facial and audio recognition to activate a privacy shutter when recognised household members arrive home.
AI has enormous potential to keep us safer by making the protection of people and assets more efficient and cost-effective. It will free humans from mundane, repetitive tasks, making our jobs simpler, our efforts more calculated and our outputs more accurate.
However, it will be up to we humans to ultimately determine and enforce the acceptable limits of its reach.