LAUNCH OF THE AUSTRALIAN STRATEGIC POLICY INSTITUTE’S BIG DATA IN NATIONAL SECURITY REPORT

 

Thank you for your kind introduction. I acknowledge that we are here tonight on Aboriginal land, and I acknowledge Indigenous elders past and present including any people of Indigenous descent with us this evening. 

 

I was delighted to be asked by Dr John Coyne to come to ASPI tonight to launch the report, Big Data in National Security, by Michael Chi.

 

It is a fascinating report. And – like all of ASPI’s work – it will no doubt broaden public knowledge and help shape the debate surrounding the use of big data in law enforcement and national security.

There have been some big changes to the law enforcement landscape over the past fifty years, but I think it’s arguable that none are as important as the growth of online activity.

 

It’s believed the first legal sale online occurred sometime in 1994.

 

There is a debate about the very first item that was bought online. One popular view is that Pizza Hut scored that first sale – a pizza sold through its old ordering website, PizzaNet.

 

But Tom Wainwright of The Economist has found that the first illegal online sale occurred more than twenty years before this large pepperoni was legally transacted.

 

It was between bunch of students at Stanford and MIT who sold a bag of marijuana through Arpanet.

 

People using the new technology for nefarious purposes had an almost-20-year head start.

 

And just as online shopping has grown exponentially since that time, so too has the criminal use of online technology.

 

Only a few years ago, the Silk Road online marketplace was shut down. According to the FBI, Silk Road did $1.2 billion worth of business between February 2011 and July 2013. That’s the equivalent of sales in a pretty large Australian company. It operated in plain sight, facilitating criminal activities hundreds of times a day. And it took two years to shut it down.

 

I begin with these tales because they are the most familiar starting point for a conversation about the set of assets and technologies that are the subject of the report being launched this evening.

 

In my role as Shadow Minister for Justice, I hear from law enforcement across the country about the new challenges they face in the digital era. The Internet is being used to buy and sell guns, engage in fraud and plan terrorist attacks.

 

Indeed, big data is being used extensively by criminal networks, to improve the hit rate on phishing, to profile targets for fraud and exploitation by attempting to steal vast amounts of tax or bank data.

 

This type of technology has been a game-changer for criminals and terrorists. This is well known.

 

But tonight, ASPI is asking us to have a conversation with a different tone, to speak about possibilities rather than simply threats.

 

And I’m glad for this, because the possibilities on the positive side of the ledger are exciting.

The report looks specifically at the question of how big data, machine learning and the associated analytics could be used in a national security frame, and I know Michael will speak to these in detail.

 

But as Shadow Minister for Justice, I can see a lot of significant criminal behaviour that we can tackle through the smart use of these technologies.

 

Take for example, the issue of sex trafficking – surely one of the most barbaric and inhumane crimes.

 

We know that sex traffickers advertise online, often publicly and through social media, but using ever-changing codes to hide their conduct. Machine learning, where computers can learn as they go, improving their capability without being reprogrammed, is proving quite effective in helping analyse not just plain words and pictures in volumes of data, but concepts as well as underlying or hidden meanings.

 

Analytics could be used in this context to predict future criminal behaviour. We think about this often in the context of law enforcement – so called ‘predictive policing’ – but there’s also the potential here to use big data to generate warnings for potential victims of human trafficking.

 

Indeed, even examining things like aberrant rates of energy use between buildings could assist in finding victims who are trapped in slavery here in Australia.

 

This example shows how these new technologies can be harnessed for the common good.

I want to make some comments on the policy issues. Big data and the internet are pervasive and transformative developments, so they throw up many issues. I want to mention three.

 

First, we need to be aware that we do not live in a one model world.

 

Nobel Prize-winning economist Thomas Schelling once said “there is a tendency in our planning to confuse the unfamiliar with the improbable”.

 

Take, for example, the 9/11 attacks. Planes being used as missiles to take out the Twin Towers was unfamiliar. Nothing like it had ever happened before. Donald Rumsfeld in his memoir called the attack an “unknown unknown”.

 

But tragically it did happen. And it highlighted how important it is to consider unknown situations. Indeed, the 9/11 Commission found that “the most important failure was one of imagination”.

 

By their nature, big data and the associated analytics usually use the past to predict the future. But perfect prediction is impossible. If we are going to use big data to predict crime or security threats, we must continue to keep an open mind to possibilities that don’t fit our model.

 

And this leads me to my second point – we must be alive to the possibility of mistakes.

 

Some conclusions from big data analytics will be wrong. This is the case with all technology.

 

Take, for example, what we have learned from DNA evidence.

 

Over the past three decades, a series of cases have highlighted how human error or even technological errors can lead to innocent people being convicted of crimes they did not commit.

 

Overreliance on big data analysis may also lead us to make catastrophic errors.

 

It can mean that innocent people are wrongly targeted by law enforcement.

 

It can mean that those who seek to do us harm go undetected.

 

As the report acknowledges, we must make sure the way that we use big data and analytics addresses the potential for false positives and false negatives – because our models and algorithms will not always be right.

Finally, we must ensure public confidence in law enforcement’s use of big data and analytics.

 

There are big ethical issues at stake here, and we cannot ignore them.

 

I believe there is a need for a more thoughtful and robust public conversation about privacy in particular. It is a conversation that flares up over specific issues – most recently, over telecommunications interception and surveillance.

 

We should and we must engage with those concerns seriously and in good faith. We must make sure they are properly addressed and that appropriate protections are in place.

 

It is an old debate. But it is not likely to quieten down in future years.

 

The lesson here, I believe, is that law enforcement agencies and policy makers must work together to ensure there are robust protections in place for people’s rights and privacy.

 

The report makes a number of recommendations about how to ensure we do this.

 

I am delighted that ASPI has undertaken this report, and I congratulate Michael Chi on his contribution to the debate.

 

It couldn’t come at a better time.

 

Criminals and terrorists are exploiting the Internet to plan and carry out criminal acts.

 

We need to take the fight online, including through the use of big data and analytics.

 

There won't be a single solution. There won’t be a simple solution.

 

But we come at this problem with a bipartisan commitment to keeping Australians safe.

 

And, as we will hear tonight, there is no shortage of good policy ideas.

 

I’m looking forward to speaking to you at the end and hearing more from Michael.