Summer Walker

The UN should follow the secretary general’s lead on artificial intelligence and cybercrime.

There is no doubt that artificial intelligence is rapidly becoming part of everyday life. In the public sphere, the authorities are increasingly using facial-recognition technology to identify people; in our private lives, mapping and translation apps are now commonplace; and machine learning is applied in fields as diverse as conservation, healthcare and food cultivation.

As AI becomes more powerful and wide-reaching, many at the international level are debating the dual nature of AI technology – in other words, its potential to beneficially transform society and foster development, while also introducing new risks to global security.

At the UN, AI and the Fourth Industrial Revolution are among Secretary General Guterres’s 2018 key priorities. He has proposed capitalizing on the UN’s role as an international forum to push forward discussions on the topic among member states, the private sector, scientists and civil society. Conversations about AI have been initiated in a number of important areas, including disarmament and development.

In terms of the potential beneficial applications of AI for development – particularly in relation to Agenda 2030 – there has been some momentum. In May, the UN’s International Telecommunications Union (ITU) hosted its second AI for Good Global Summit, which focused on AI applications to advance the Sustainable Development Goals. Related to this, the ITU published a special journal on the potential positive contributions of AI. The UN’s Chief Executives Board, the organization’s highest-level coordination body, is also working on AI. In April, its High-Level Committee on Programs approved a process to improve UN engagement on AI capacity building related to Agenda 2030 and asked the ITU to draft a system-wide framework for this.

Regarding disarmament, the secretary general’s recently published agenda includes sections on cyberspace and artificial intelligence, raising concerns that game-changing technologies are emerging with greater frequency than ever before, and that their autonomous attributes have the ability to lower the threshold of use of force (as customarily understood in the UN charter and international law more widely). He also warns that the pervasiveness of these technologies could increase the risks, range and impacts of attacks, including those by non-state actors. Guterres encourages states to coordinate without delay if they hope to have a preventative impact. One state-led body working on this is the Group of Governmental Experts on emerging technologies in the area of lethal autonomous weapons systems within the Convention on Certain Conventional Weapons framework.

Cybercrime: The growing threat from AI

The secretary general has also raised the issue within the UN’s crime corridors. At the annual meeting of the Commission on Crime Prevention and Criminal Justice (CCPCJ), Guterres drew attention to how AI – and other new technologies associated with the Fourth Industrial Revolution – has the potential to create new forms of crime. However, the recent reports from this meeting and the expert group on cybercrime do not address AI, suggesting the issue has not yet made it into the state-led UN cybercrime discussions. Indeed, the potential impact of AI on cybercrime is one area where attention to the subject has arguably not kept pace with the potential threat.

There are, of course, some UN crime-related entities working on AI. In 2017, its Interregional Crime and Justice Research Institute (UNICRI) established a Centre for Artificial Intelligence and Robotics. Its objective is to educate policymakers on the risks and benefits of AI and robotics to inform policymaking. It recently held a meeting with INTERPOL where cybercrime was part of a larger discussion about the risks posed by criminal use of AI and potential responses using AI technology. Generally, though, the issue seems not to have gained the same momentum in the cybercrime sphere as it has in development and disarmament circles.

Yet current trends suggest that AI should assume a more central role in cybercrime discussions. Like its potential to have a positive transformative effect, AI also has the potential to transform how crime, and cybercrime in particular, is conducted. With AI, criminal actors will be able to hone their ability to carry out attacks and find new opportunities for attacks, such as altering the signalling system in driverless cars. AI could automate time-consuming tasks, such as spear phishing, to reach a larger target audience more quickly. Physical security could be at risk by creating or hacking automated weapons systems. And, as more personal data is collected and sifted by means of algorithms, criminal actors will have new ways to corrupt political processes or extort individuals based on their personal data.

The increasing pervasiveness of the Internet of Things (IoT) adds to the potential for new, possibly more personal, forms of cybercrime. Key trends in cybercrime identified by the chief of the UNODC’s Global Cybercrime Program include increased use of IoT in cybercrime and ransomware attacks, both of which could be enhanced through AI technologies.

The proprietary nature of AI is also changing, with more open-source coding and training available to a growing number of users – examples being Google’s open-source machine learning initiative, TensorFlow, and the Shenzhen Open Innovation Lab. While these advances offer potential benefits in many developing contexts, from food cultivation to healthcare systems, they also need to be considered in light of the risk of criminal activity. Such developments could provide criminal actors with easier access to AI tools, while making it more difficult to regulate how AI is used.

There are also potential positive applications for AI in addressing cybercrime. AI could increase the reach for detecting and defending against cyber attacks, some of which have had global reach. At one point, Google claimed it stopped 99% of incoming spam using its machine learning technology. Some also believe AI could become a useful tool to help attribute attacks to perpetrators – whether a criminal act carried out by an individual or a security breach by another state.

In many ways, AI has become a constant in many people’s lives, yet it remains somewhat of a mystery how it might impact policy areas. This ambiguity has not stopped UN members from engaging in debates on the subject in areas such as development and disarmament. As cybercrime discussions progress within the UN, paying attention to AI’s potential as a criminal tool could help states better understand and respond to the potential consequences.