South Africa withdraws AI policy after fake AI-generated citations discovered

South Africa has pulled back its proposed national AI policy because parts of it were written by artificial intelligence and included references to made-up sources.

On April 10th, South Africa released a proposed national policy for Artificial Intelligence, aiming to position the country as a hub for AI development. The policy outlines how South Africa plans to tackle the societal, economic, and ethical issues related to AI, and details the creation of an AI commission, an ethics board, and a regulatory body.

South Africa’s Communications Minister, Solly Malatsi, pulled the policy after News24 revealed that at least six of its 67 citations were actually made-up by artificial intelligence – despite the document appearing to be based on legitimate academic research.

The aftermath of the failure

Each of the journals mentioned in the proposed policy confirmed that the articles it cited were fabricated, and they did so independently of one another.

In a post on X (formerly Twitter), Malatsi explained what went wrong. He stated the most likely cause was the inclusion of citations created by AI, which weren’t properly checked. He emphasized this wasn’t just a technical mistake, but a serious issue that damaged the trustworthiness of the proposed policy.

He explained that this mistake highlights the crucial need for people to carefully monitor how artificial intelligence is used. He added that they are learning from this experience with a sense of humility. According to Malatsi, the people who wrote the flawed content will be held accountable.

It is expected to be corrected and reissued for public comment.

Read More

2026-05-02 15:18