Can AI finally stop corruption in South Africa?

Artificial intelligence is being harnessed to help weed out corruption by creating an early warning system based on data processing.  

That’s according to the Institute for Security Studies, which is working on an AI tool currently at an early stage of development.   

Criminals who may have outsmarted traditional criminal intelligence may find AI a more formidable foe, with similar initiatives already well-established abroad.

“The tool we are building is an early warning system for detecting state capture,” comments project leader Nicola Prins.

“It will be focused on diagnosing whether the patterns that have been found in state capture are presenting themselves at government institutions. The system is making use of AI, to help collect and analyse data, but is human led.”  

She said measurable indicators were incorporated into to system to help identify whether state capture is present in different contexts.

“The system doesn’t try to quantify the extent of corruption; i.e. number of cases or the values involved. Our focus on identifying which elements of state capture are present will help inform what action can be taken for each individual institution, with the aim of preventing corruption and preventing capture from becoming more entrenched in the institution,” Prins said.  

The ISS tool focuses on prevention and early intervention. It recognises that that while criminal justice system can only work in a reactive manner, other agencies within the administrative domains of government are in a position to intervene early to disrupt malfeasance.  

The National Anti-Corruption Advisory Council (NACAC) has recommended that Cabinet approve a permanent, independent anti-corruption body known as the Office of Public Integrity and Anti-Corruption. The recommendation is in part a response to the State Capture Commission.  

The NACAC report also recommends the use of Artificial Intelligence to prevent corruption.   

President Cyril Ramaphosa this week acknowledged the urgent need to address the NACAC’s concerns:  “Corruption is much more than the result of the criminal intent of a few individuals be they government officials, elected public representatives or business people,”  Ramaphosa said in his weekly online newsletter.

“It can become embedded in state institutions or business enterprises and manifest itself in practices and organisational culture. The success of our efforts relies on our ability to prevent corruption in the first place in state institutions, business enterprises or organs of civil society.”  

“That is why we need to build transparent, accountable and ethical institutions – both public and private – in which corruption is unable to take root. We all need to work together to build a society characterised by responsibility and integrity,” Ramaphosa said.  

Colette Ashton, a research consultant within the ISS Anti-Corruption and Accountability project, said corruption was notoriously difficult to quantify, hence the need for analytic tools: “Corruption, as a secretive activity, is extremely difficult to measure. This is why the major indices, such as the Corruption Perceptions Index, rely on data about peoples’ perceptions of corruption, rather than data about corruption itself.”  

Around the world, the UN, OECD and International Anti-Corruption Academy are working on a project to start trying to source and measure this data in more accurate ways. These measurements are essential for formulating policy. Yet, because most corruption is hidden, it is difficult to measure the quantity of corrupt activity, let alone the opportunity cost thereof,” Ashton said.