What happened
Singapore's Personal Data Protection Commission (PDPC), under new Commissioner Denise Wong, proposed draft advisory guidelines (reported July 13-15, 2026) requiring organisations to issue AI-specific notifications — rather than relying on broad 'new product development' privacy notices — when personal data is used to train generative AI models. The guidelines would require disclosure of what the AI model does, what data types are used, and clear opt-out/consent-withdrawal mechanisms. Key open questions remain, including whether the requirement applies to anonymised training data.
Why it matters
This tightens transparency obligations under Singapore's PDPA specifically for the generative AI training pipeline, affecting any business — local or foreign — that trains or fine-tunes generative AI models using data from Singapore users, including voice, image, and text data collectors (e.g., text-to-speech, social media platforms).
Action needed
Businesses training or fine-tuning generative AI models using Singapore user data should review current privacy notices and prepare to implement AI-specific notification flows and opt-out mechanisms in anticipation of finalized guidelines.