|
Abstract :
|
As AI applications and services iterate rapidly, consumers' right to be informed is expanding from the narrow question of whether personal data are collected to a continuous, actionable and remedy-oriented understanding of key facts-whether content is generated or synthetically produced, whether services are driven by algorithmic recommendation or automated decision-making, whether interaction data are reused for training and profiling across scenarios, and where the boundaries lie in anthropomorphic interactions. In recent years, alongside advancing its "AI+" strategy, China has built on the Personal Information Protection Law as a baseline and adopted scenario-based measures on recommendation algorithms, deep synthesis, generative AI and content marking. It has also promoted a shift from text-based compliance to closed-loop, mechanism-based governance through structured list-based disclosures, coordinated responsibilities across multiple actors in the value chain, and compliance audits-resulting in a set of core mechanisms, including layered notice, structured lists and SDK transparency, opt-out personalization with verifiable effectiveness, controllable profiling tags, explainable and reviewable high-impact decisions, dual-track explicit/implicit marking, dynamic prompts and exit options for anthropomorphic interactions, and evidence-chain recordkeeping.
|
Word
