The Ministry of Digital Development and Information (MDDI) is implementing guidelines to enhance artificial intelligence (AI) governance in Singapore, aiming to create a more business-friendly AI ecosystem. During her opening address at Personal Data Protection Week, Minister for Digital Development and Information Josephine Teo highlighted the necessity of mitigating data risks to foster trust in AI-enabled products and services among businesses and consumers.
“In the age of AI, the need for data and trust has come into sharper focus. Today’s generative AI-enabled products and services, such as chatbot applications, are being built on the foundation of more and more powerful large language models, or LLMs”, she said.
She remarked that businesses will still require data to deploy applications on top of existing large language models (LLMs) for the foreseeable future. However, she pointed out the risks associated with the datasets used in AI development, such as the potential for bias in the results generated by these models. Since these datasets may include personally identifiable information, there is a risk that generative AI models built on them could inadvertently reveal such information when prompted.
“If these risks are not mitigated, businesses and consumers alike may find it difficult to trust AI-enabled products and services. Without a foundation of trust, support for AI innovations could diminish over time”, said Teo.
With this in mind, MDDI plans to introduce a set of safety guidelines – which will be part of a testing framework and software toolkit AI Verify – for generative AI model developers and app deployers. The guidelines seek to establish a baseline, common standard for generative AI development with a focus on transparency and testing. The Infocomm Media Development Authority (IMDA) will start consulting the industry on these guidelines to ensure they are “relevant and robust”, said Teo.
MDDI is also aiming to promote greater adoption of privacy-enhancing technologies (PETs) in AI, as these technologies can enable businesses to leverage data more effectively while safeguarding personal information. The Personal Data Protection Commission, through IMDA's PETs sandbox, has introduced a proposed guide on synthetic data generation. This guide is designed to assist businesses in understanding synthetic data, including its definition, potential applications, and best practices for its creation.
Teo further announced that a new Asean guide on data anonymisation will be released early next year. “I hope it will serve as an important and practical resource for businesses in Asean looking to anonymise data for greater and more responsible use of data across the region”, she said.