Department of Home Affairs minister Dr Leon Schreiber. (Photograph by DHA)
Home affairs minister Dr Leon Schreiber will tomorrow demonstrate the new Electronic Travel Authorisation (ETA) system, as he continues his fight against corruption by deploying artificial intelligence (AI).
The Department of Home Affairs (DHA) is prioritising anti-corruption efforts through digitisation and AI, which are key to cleaning up the entity and improving efficiency to collect more revenue.
Schreiber cites a recent issue in which 95 Libyan nationals snuck into the country due to loopholes in the system that allowed them to collude with DHA officials to get falsified documents.
He asks how it is “possible that in the year 2024, in the age of AI and self-driving cars, the most developed country on the African continent still used pens and pieces of paper that can be obviously manipulated, to grant access to our country”.
To resolve such issues, Schreiber says systematic reform to weed out corruption will largely be based on digital transformation.
The ETA, a part of the DHA’s arsenal, is under development and will be phased in. It will eventually automate South Africa’s visa processes.
“It will use the same principles successfully used by the South African Revenue Service to completely automate not only the application process, but also the adjudication process for visas,” Schreiber says.
The tax office makes use of AI to match taxpayer-provided data with information from third-party sources such as banks, medical aids and employers, among others.
Schreiber explains that visa applicants will need to apply online, scan their passport, take a selfie and provide other required information.
“Through machine learning technology, the ETA will check 40 different features of their passport to ensure it is authentic, it will match their selfie to the photo on their passport, and when they arrive at a port of entry, a camera will confirm that the face of the traveller who arrived is the same as the one who applied,” says Schreiber.
This will eliminate the scope for human interference, bias and corruption, the minister says.
“In the first phase, we will roll out the ETA for simpler applications like tourist visas shorter than 90 days. But over time, our ambition is to roll it out for every visa category,” Schreiber says.
“The launch of the ETA, set for the end of September 2025, will boost tourism and investment, making South Africa more competitive, alleviating burdensome red tape when applying for a visa,” Investec economist Lara Hodes has said.
The department has also begun implementing a similar technological upgrade that will lead to major improvements to South African passport security, Schreiber says.
He explains that a “systematic weakness” that allowed “syndicates” to exploit a lack of biometric verifications at the start of the application process is being closed.
“If the face on that photo does not match the face we have on record in the population register, the application is declined and the passport is not produced,” says Schreiber.
Another aspect of digital solutions is the previously announced enhanced partnership with banks to offer South Africans the ability to apply for smart IDs in their branches, with the longer-term vision being enabling such applications via banking apps.
Capitec, First National Bank, Absa, TymeBank and Standard Bank are among those that have signed up to provide this solution.
However, AI solutions are themselves subject to potential fraud. A TransUnion report on the rapid rise of AI-powered fraud and deepfake scams found that deepfake-linked fraud surged globally last year.
TransUnion says audio and video deepfakes accounted for 7% of all fraud incidents. “Cases of audio-based impersonation scams rose from 37% to 49% year-over-year and video deepfakes doubled, fuelling identity theft, investment scams and high-value authorised push payment fraud.”
Amritha Reddy, senior director of fraud product management at TransUnion Africa, says the issue is that traditional fraud detection was designed for an earlier era. Fraud has always been a moving target, she says.
“What makes the current wave so alarming is the pace, scale and realism of AI-powered deception. Bots operate around the clock, processing in milliseconds what used to take fraudsters days.”
Reddy adds that “fake voices, synthetic documents, and fabricated videos are now often indistinguishable from reality, placing businesses, consumers and governments under constant siege”.
The only viable path forward is to fight AI with AI, says Reddy. “Just as fraudsters harness machine learning to deceive, businesses must deploy AI-powered defences to protect.”
