IMDA Generative AI Chatbot Transparency Guidelines

On 20 July 2026, the Infocomm Media Development Authority (IMDA) launched Generative AI Chatbot Transparency Guidelines at the Singapore Data Festival, framed as a first step toward accountability at the application layer.

The problem being solved. For most people, the generative AI application they meet most often is the chatbot. We use it, but often do not know its limitations, or what happens to our data. The information usually exists — it is just scattered across terms of service, privacy notices and other documents, and is typically either too simplistic or too technical for ordinary users.

The approach: an information card. The guidelines call for a Chatbot Information Card that works like the label on a medicine package — it does not list every scientific detail, but presents the things a user actually needs to know in one readable place: what this bot is for, what its limitations are, how your data is handled, and where you can turn when something goes wrong.

Division of labour with the PDPC guidelines. The PDPC's Advisory Guidelines on the Use of Personal Data in Generative AI, issued the same day, govern the data layer — the notice and consent obligations organisations owe when training models on personal data. IMDA's guidelines govern the application layer — whether an end user facing a chatbot can understand what they are using. Together they form Singapore's two-layer accountability design for generative AI.

It also continues Singapore's consistent pattern: start with a voluntary, operational, toolised framework (AI Verify and the Model AI Governance Framework both followed this route), let industry move first, rather than legislating in one step.