Absa has warned that customers who give an AI agent their banking login could be held liable for any losses that follow. FNB, Capitec and Discovery Bank said they do not allow outside AI agents into customer accounts.
The banks gave their views to TechCentral. This came after Axios reported that big US banks had found unauthorised AI agents in customer accounts. Some of those agents appeared to act without the user telling them to.
“Where a customer voluntarily shares such information and a financial loss subsequently occurs, the customer may be held liable,” said Jochem Melis, CIO of Absa’s engineering, services and platforms office.
Melis said Absa does not use AI agents to access customer accounts. He said the bank strongly discourages customers from sharing their banking details with any outside AI agent.
He said the agents cannot get into accounts on their own. Access is only possible when a customer hands over the login details or permissions. “From a security perspective, providing banking credentials to an AI agent is no different from sharing them with any other third party,” Melis said.
Absa said agents do not change its current risk profile. Melis said its fraud systems are built to spot suspicious patterns, including attempts at unauthorised agent access. Losses will be assessed under the Absa digital fraud warranty.
“FNB does not currently allow third-party AI agents to access customer accounts or make transactions for customers,” said Dané Liebenberg, AI executive for retail and business banking at FNB.
Liebenberg said FNB is building a secure way for approved AI assistants to connect to the bank in future. Customers would give consent for every action. Identity checks would always happen on FNB’s own platforms, and every interaction would be recorded.
The bank will start with information services and expand only once the right protections are in place. “Our principle is simple: no autonomy without control,” Liebenberg said.
Capitec said it does not allow outside AI agents into client accounts, and that this includes screen-scraping tools. “A client’s account can only be accessed by the client, through Capitec’s own secure channels,” the bank said.
Capitec told clients not to share their Pin, password or one-time Pin with any third party, including AI tools. The bank said it is not yet offering AI products directly to clients.
However, group CIO Andrew Baker wrote on 7 October that Capitec was rolling out Scam Checker. He described it as a free AI tool inside the Capitec app for checking suspicious links, texts and images.
Discovery Bank said it does not allow outside AI agents to access client accounts or start transactions. It is building its own tools through Discovery AI instead.
Discovery AI lets clients prepare a payment in the app using plain language or a photo of an invoice. The client must check and approve the payment before it goes through.
All four banks said they would look at any losses case by case. FNB said it would follow its terms and conditions and the relevant regulations.
Standard Bank and Nedbank had not answered TechCentral’s questions by the time of publication.
In the US, JPMorgan Chase CEO Jamie Dimon said on 13 October that his bank was “not going to allow just any outside agent” in. Axios said Morgan Stanley has started blocking outside agents.
Visa has signed up South African banks to its Agentic Ready programme to prepare them for AI shopping agents. On 15 October, the Financial Stability Board finalised its guidance on AI in financial institutions, which calls for human oversight.
Source: this article is based on reporting by TechCentral. Image: TechCentral. Written with the help of AI and published by the Tzaneen Voice Technology desk. See our Editorial Standards.
Werner Jacobs covers technology for Tzaneen Voice. Werner reports on mobile networks, data prices, internet access, smartphones, apps, AI, cybersecurity and South African tech companies. Each story explains what a new product, price change or policy means for ordinary users in South Africa.