AI Regulation Moves From Principles to Compliance as Deepfake and Data Rules Tighten

Singapore, 20 September 2026 – Artificial-intelligence regulation is entering an enforcement phase as governments move beyond voluntary principles toward rules covering deepfakes, algorithmic safety, training data and personal privacy. The shift is creating a more demanding compliance environment for developers and corporate users, with obligations increasingly attached to how models are trained, deployed, monitored and explained.

The European Union’s framework now applies transparency duties to certain interactive and generative systems. Users must be told when they are interacting with AI, while synthetic content and deepfakes face labelling requirements. Enforcement is divided among national authorities, a central AI office and the data-protection supervisor for EU institutions, creating a multi-layer system that companies must map carefully.