South Africa has withdrawn its first draft national artificial intelligence policy after discovering its reference list contained fictitious sources that appeared to be AI-generated.
The withdrawal marks an embarrassing setback for the country's efforts to establish formal AI governance frameworks. The draft policy, intended to guide South Africa's approach to artificial intelligence development and deployment, included citations that did not exist and appeared to have been fabricated by an AI system.
The discovery highlights a growing concern in policy-making circles: the risks of relying on AI tools without adequate verification processes. As governments worldwide rush to develop AI regulations, the incident underscores the importance of rigorous fact-checking and human oversight in policy documentation.
South Africa's Department of Communications and Digital Technologies had been working on the national AI policy to address opportunities and risks associated with artificial intelligence adoption. The framework was expected to cover areas including innovation, workforce development, ethical considerations, and regulatory approaches.
The use of AI-generated fictitious sources in a policy document about AI governance created particular optics challenges, suggesting potential gaps in the review process. It remains unclear at what stage the false citations were introduced or whether they resulted from prompts given to AI systems or inadequate verification of automatically generated content.
The withdrawal signals that South Africa will need to restart its policy development process with enhanced quality control measures. This includes implementing more stringent verification protocols for all source materials and citations before publication.
Other nations developing AI policies are likely to view this incident as a cautionary tale. As policymakers increasingly use AI tools to assist in research and drafting, the need for comprehensive human review and fact-checking has become more critical.
South Africa has not yet announced a timeline for releasing a revised version of its national AI policy.
Z.ai released GLM-5.3's weights on Hugging Face under a new license that requires large companies to undergo security review before hosting the model. The change marks a departure from the standard MIT license.
Anthropic has introduced the Model Hardware Standard (MHS), a unified interface enabling AI agents to operate robotic arms, lab instruments, and other physical devices. Early testing shows integration time has dropped from weeks to hours.
Open-weight AI companies—those releasing freely available models—are attracting major acquisition interest from tech giants. The trend reflects growing capital investment in the business model of distributing AI models at no cost.
Google Deepmind has upgraded its Co-Scientist AI system to autonomously plan experiments, operate lab equipment, and publish scientific papers. The Gemini-based multi-agent platform demonstrated experimentally validated results across materials science, chemistry, and medical AI development.