Singapore – The Infocomm Media Development Authority (IMDA) has published a set of voluntary transparency guidelines aimed at helping consumers use generative artificial intelligence (GenAI) chatbots more safely and responsibly, as the technology becomes increasingly integrated into daily life.
The Transparency Guidelines for GenAI Chatbots, which IMDA describes as among the first of their kind globally, seek to standardise how chatbot developers and deployers communicate information about their AI applications to the public. The authority said greater transparency can help users better understand what chatbots can and cannot do, as well as how potential risks are managed.
According to IMDA, the guidelines are intended to support informed use of AI chatbots by enabling users to understand an application’s capabilities and limitations, verify outputs where appropriate, and recognise situations where chatbot use may not be suitable.
The framework also aims to standardise disclosures across chatbot providers, allowing consumers to compare services more easily and locate relevant information without navigating multiple sources. In addition, it seeks to enhance accountability by encouraging deployers to publicly outline their safety commitments and policies.
A central feature of the guidelines is the introduction of a chatbot information card, which may take the form of a dedicated webpage or disclosure document. Similar to a nutrition or medicine label, the information card is designed to provide users with essential details about a chatbot in one place.
The information card is built around three principles: relevance, accessibility and timeliness. It is intended to explain what a chatbot can and cannot do, its reliability and safety measures, how user data is handled, and how users can report issues or concerns. IMDA recommends that the information be written in plain language, be easy to locate and navigate, and be updated whenever significant changes affect a chatbot’s capabilities, risks or safety policies.
The guidelines were developed through consultations with technology companies, enterprises operating consumer-facing chatbots, government agencies and local users. IMDA said the framework was designed to balance consumer expectations with practical implementation considerations for chatbot deployers.
Several organisations have indicated that they will use the guidelines as a reference over the next 12 months to improve transparency practices for their public-facing chatbots. These include global AI companies such as Google for its Gemini app and Meta, as well as Singapore-based organisations including DBS, OCBC, Singapore Airlines and Synapxe.
Sector regulators have also recognised the initiative. IMDA said the Monetary Authority of Singapore’s proposed guidelines on artificial intelligence risk management and the Ministry of Health’s updated Artificial Intelligence in Healthcare Guidelines both emphasise transparency as a key principle. The new chatbot transparency guidelines are intended to complement these sector-specific frameworks where GenAI chatbots are deployed.
Singapore’s public sector agencies, including the National Library Board and the Health Promotion Board, also plan to reference the guidelines when providing information about their own public-facing chatbots.
IMDA said the guidelines initially focus on GenAI chatbots because of their widespread consumer use and increasing concerns surrounding issues such as data privacy, child safety and risks to mentally vulnerable users. However, it noted that the framework can also serve as a reference for other GenAI applications, while sector regulators may build upon it to develop more specialised transparency guidance for their respective industries.
The guidelines are part of Singapore’s broader AI governance approach, which seeks to balance innovation with safeguards that promote public trust. IMDA said it will continue working with sector regulators and international organisations to extend the principles to other AI applications and contribute to global efforts on AI transparency by shifting attention from model-level to application-level transparency.
