Examine UNESCO's Recommendation on the Ethics of Artificial Intelligence (2021) and its relevance to safeguarding press freedom and media integrity in the AI era.

Adopted in November 2021 by UNESCO's 193 member states, the Recommendation is the first global normative instrument on AI ethics [1]. Its value for the media lies less in prohibition than in placing human accountability at the centre of machine-assisted publishing.

What the Recommendation contains

• Four core values: human dignity and rights, just and peaceful societies, diversity and inclusion, and environmental flourishing [1].
• Ten principles, including proportionality, transparency and explainability, accountability, fairness and human oversight [1].
• Actionable policy recommendations across eleven areas, taking it beyond a declaration of intent — though it remains soft law, dependent on state implementation [1].

Relevance to press freedom and media integrity

• Transparency and labelling: UNESCO, with Reporters Without Borders and the Journalism Trust Initiative, translated these principles into newsroom asks — clear identification of machine-generated content and retained human editorial judgment [2].
• Accountability stays human: editors increasingly lean on automated AI-detection tools, but the framework insists responsibility cannot be outsourced to a probabilistic score [2].
• Guarding against bias: UNESCO warns that discriminatory training data makes discrimination "systematic and automatic" [2] — a live risk where second-language contributors are wrongly flagged as machine writers.
• Self-regulation route: press councils across South-East Europe and Türkiye adopted AI-ethics standards, with Serbia revising its Code of Ethics — a model the Press Council of India could adapt [2].

Gaps in the Indian context

• The India AI Governance Guidelines (2025) embed comparable sutras — trust, accountability, understandable by design — but are guidance, not statute [3].
• The IT Rules amendments define synthetically generated information and mandate labelling with minimum 10% visual coverage, binding intermediaries and significant social media intermediaries [4] — designed for deepfake video, leaving AI-written text largely uncovered.

The Recommendation therefore supplies the ethical vocabulary India's media governance still lacks in enforceable form. Embedding disclosure-at-authorship, a right of reply for flagged contributors, and a Press Council code would convert these global norms into workable safeguards for both press freedom and public trust.