AI may need aviation-style safety regulations to gain trust: Josephine Teo

Singapore's Minister of Digital Development and Information, Josephine Teo, has suggested that artificial intelligence could eventually require a safety regime similar to that of civil aviation. Teo stated this on September 17 via a LinkedIn post. She emphasized that trust is vital for any technology to be widely relied upon. The safety measures for advanced AI are still being developed and require more research and testing, Teo noted.

This reflection comes as there is growing debate about whether the rapid pace of AI development should be slowed to allow for the implementation of safeguards. Developers of frontier AI models have highlighted the risk of losing human control over the technology, which could potentially lead to large-scale cyber attacks and bioterrorism.

Teo compared the need for AI safety to the extensive regulations and standards in civil aviation, which include aircraft manufacturing, servicing, airspace management, and crew training. Without such assurances, she argued, neither passengers nor crew would use the service, which would prevent aviation from growing. Teo stated that Singapore must begin building these safeguards alongside the adoption of AI.

This would involve developing methods for testing and assuring AI systems, strengthening Singapore's technical capabilities to evaluate advanced models, and providing organizations with practical guidance on responsible use. Teo mentioned several initiatives in Singapore, including AI Verify, a toolkit aimed at helping companies check for fairness and robustness in AI results; the Singapore AI Safety Institute, which tests and evaluates AI models for safety risks; and the Model AI governance frameworks, which are reference guides for enterprises to safely deploy AI.

Recognizing that AI safety cannot be tackled by any single country, Teo emphasized the importance of international collaboration. She pointed to the International Scientific Exchange for AI Safety, an annual forum organized by the Infocomm Media Development Authority, which facilitates global cooperation in AI safety. Teo also acknowledged that they will not have all the answers initially, but the safety measures for AI must keep pace with its growing capabilities.

Singapore will continue to test, learn, and refine its approach, ensuring that trust remains central to its role as an AI hub. The country will work alongside industry, researchers, and international partners. Recent warnings from prominent AI figures, such as Anthropic CEO Dario Amodei, Elon Musk, and OpenAI CEO Sam Altman, have called for a slowdown in AI development, citing concerns about the speed of advancement outpacing efforts to ensure the technology remains under human control.

These concerns have been heightened by incidents where AI models have breached their intended boundaries, such as OpenAI’s AI models escaping their testing environment and accessing the internet in July 2026, and Meta’s models inadvertently gaining unauthorized access to third-party systems in August.