In May 2018 a study undertaken by global research giants, Frost & Sullivan, and commissioned by Microsoft, revealed that the potential economic loss across Asia Pacific in 2017 due to cybersecurity incidents could have hit a whopping US$1.745 trillion. This is more than seven percent of the region’s total GDP (US$24.3 trillion) and a figure so big it’s difficult to comprehend.
Closer to home, the same report said that the economic loss as a result of cybercrime to Australia could be as much as $29 billion per year, the equivalent of 1.9% of the country’s entire GDP. For an Australian organisation that has more than 500 people working in it, a direct loss from a cyberattack could cost the company over $30 million.
On the other end of the scale, according to a report by cybersecurity software, Norton, over half a million small businesses fall victim to cybercrime every year with the average cost to a small business being at a staggering $1.9 million.
Of the Australian respondents to Norton’s research, over half of them said they had experienced a cybercrime incident in the past five months, however many victims of cybercrime never report it in an attempt to mitigate indirect losses such as reputational damage, so the figure could be much higher.
A new hope
A second approach in AI is to use supervised algorithms (algorithms that it has been trained on) to detect threats. Thousands of examples of malware code is provided and even if the malware mutates, part of the code will remain and will be picked up by the AI program.
Of course humans are still required to decide which action to take and how best to protect the business through integrated decision making. Humans still do a better job of prioritising actions, using common sense and seeing the bigger picture that decisions are made in.
Meanwhile, advances in deep learning – a step beyond machine learning – allows AI to mimic the working of the human brain to assist AI to reason better. Tech giants such as Facebook are pumping money into deep learning frameworks such as TensorFlow and PyTorch that have far wider applications than just cybersecurity, however the quantum leap that deep learning is expected to bring to cybersecurity is bound to have an effect on the economic impact of cybercrime.
Cybersecurity deep learning will soon detect and prevent any threat, then its increased prediction capabilities will become instinctive for further similar threats without any human intervention at all. Deep learning AI will offer a more sophisticated approach to security dealing with larger datasets such as hundreds of millions of malicious and legitimate files. Deep learning AI has the capacity to analyse and clarify the exact type of malware in real time – a job that usually requires a group of experts.
So is AI going to take over like in The Terminator movies? Well, artificial intelligence and human intelligence must work together for the best possible results, and without turning into an army of robots.
James Cameron’s Skynet and Arnie in sunglasses are not on their way. At least not any time soon.