South Africa recommends BRICS establish an international mechanism for the independent scientific evaluation of artificial intelligence (AI), said president Cyril Ramaphosa.
This, amid the accelerated pace of AI development, sometimes resulting in AI models escaping test environments to hack companies or blackmail people.
Ramaphosa made his comments during a session on “Shaping the Future for Inclusive Growth” at the 18th BRICS Summit in New Delhi, India.
India hosted the summit at the weekend, under the theme: “Building for Resilience, Innovation, Cooperation and Sustainability”.
Ramaphosa said AI and associated technologies have the potential to usher in a new era of productivity and human progress, and if used effectively and responsibly, it can support efforts to reduce inequality, poverty and underdevelopment.
However, he cautioned, the rest of the world must listen carefully to what those developing the most powerful AI systems are saying.
“They are warning about systems capable of advanced cyber operations, assistance in biological weapons development, autonomous action, manipulation, mass surveillance, disruption of employment and, ultimately, systems whose capabilities may become difficult for human beings to control.
“We should therefore neither fear AI nor approach it recklessly. Our responsibility is to ensure that human governance advances as rapidly as artificial intelligence itself.”
Referencing some AI industry leaders, including Anthropic CEO Dario Amodei, Ramaphosa said they agree AI companies should be subject to formal external checks, audits and government authority.
“Such oversight is standard practice in other industries whose activities have a significant impact on human safety and well-being, such as aviation, pharmaceuticals, nuclear power and financial institutions.
“We must never reach a situation where technological capability has moved beyond our capacity to understand it, supervise it and control it.”
The president said the independent scientific AI evaluation should be backed by meaningful human-control principles, mandatory reporting of serious incidents (of which there have been quite a few already), progressively stronger safeguards such as capability increases, and simultaneous investment in sovereign AI capacity for developing economy countries.
“We agree with what [People's Republic of China]president Xi Jinping has proposed on establishing an AI ecosystem for BRICS. We support continued scientific cooperation in initiatives such as the BRICS Intelligent Telescope and Data Network and the BRICS Vaccine Research and Development Centre.”
BRICS is a political and diplomatic coordination group for countries from the Global South. The group has expanded beyond Brazil, Russia, India, China and South Africa by adding five other countries that include Egypt, Ethiopia, Iran, Indonesia, Saudi Arabia and the United Arab Emirates.
The world’s leading AI companies − ChatGPT maker OpenAI and Claude developer Anthropic − have in the recent past disclosed incidents in which advanced AI models went rogue.
OpenAI’s frontier GPT-5.6 Sol, and an even more capable pre-release model that has not been publicly identified, breached their test environment and hacked open source platform Hugging Face.
Claude maker Anthropic also discovered some AI systems, as part of its stress test of 16 models, “resorted to malicious insider behaviours” including blackmail to complete a task.
Researchers from industry giant Anthropic last week warned that AI without regulation “could kill” all of humanity within the decade. Former Anthropic researcher Jacob Coxon took to social media platform X to accuse the companies (Anthropic and OpenAI) of advancing AI without acting responsibly.
He wrote: “Do not underestimate the power of this technology. These will soon be superhuman systems that can hack anything, revolutionise any field overnight, and acquire real power and resources. We have all witnessed the progress in each of these domains, and progress is not slowing.”
The latest revelations led to Amodeireportedly issuing a proposal at the weekend, calling to slow down the technology’s advancement to ensure public safety.
Similarly, in August, OpenAI CEO Sam Altman reportedly said his company would pause much of its model development for two weeks to bolster its defences, following the Hugging Face incident.
Reuters also reports that Altman told employees last week that the ChatGPT maker could pace development alongside other AI labs.
Extended collaboration
Meanwhile, China is seeking deeper cooperation with South Africa and other countries in the Global South to advance AI development, strengthen scientific capacity and shape international AI governance.
Li Xiuquan, director of the Key Science and Technology Fields Research Centre at the Institute of Scientific and Technical Information of China, said open source technologies, shared datasets and international capacity-building initiatives could make AI resources more accessible to developing countries.
Xiuquan was addressing a parallel session at the high-level conference on International Cooperation in Open Science, in Shanghai, China.
Science, technology and innovation minister professor Blade Nzimande is leading a South African delegation at the gathering. The group includes scientists, researchers, innovators and senior government officials.
Xiuquan said China viewed cooperation with countries in the Global South as an important component of its international AI programme. This includes exchanges involving Chinese researchers, United Nations institutions and participating countries.
“Through international forums, we have conducted various South–South exchange programmes involving teams from China, the United Nations and partner countries, with a focus on strengthening AI capacity. We have also conducted numerous exchange programmes under the Belt and Road Initiative, targeting the Global South,” he said.
Share
