Singapore IMDA issues voluntary genAI chatbot transparency rules

Singapore’s Infocomm Media Development Authority (IMDA) published voluntary transparency guidelines for generative AI chatbots, introducing a common framework for user disclosures.

TL;DR

• Singapore’s IMDA published voluntary transparency guidelines for generative AI chatbots.
• The framework centres on a chatbot information card with key disclosures for users.
• Providers are asked to disclose capabilities, limitations, safety measures, data collection practices and reporting channels.
• Google, Meta Platforms and Singapore Airlines plan to use the framework over the next six to 12 months.

IMDA said the guidelines are intended to standardise information for consumer AI services so users can compare applications more easily and providers can be held accountable. IMDA said the framework is voluntary, but it urged generative AI deployers to adopt at least a minimum level of transparency.

The centrepiece of the framework is a chatbot information card that IMDA described as a single reference point for essential disclosures. IMDA said providers should present the capabilities, limitations, safety measures, data collection practices and reporting channels of their products in plain language. IMDA also said the information card should be easy to access from within the chatbot interface and updated regularly as capabilities or risks change.

Google, Meta Platforms and Singapore Airlines are among the organisations planning to use the framework over the next six to 12 months. IMDA said it will monitor implementation of the information cards, refine the guidance over time, and extend the guidelines to AI applications in healthcare, education and financial services.

Related questions

• What information must Singapore’s IMDA chatbot information card include?
• Which companies plan to adopt Singapore’s genAI transparency framework?
• Is Singapore’s IMDA genAI chatbot transparency framework mandatory or voluntary?