SC asks Centre to examine proposals for regulating high-risk AI systems

The Supreme Court on Thursday asked the central government and other authorities to consider proposals for introducing binding standards governing the ethical and transparent use of artificial intelligence in surveillance and content moderation.

A bench comprising Chief Justice of India Surya Kant and justices Joymalya Bagchi and V. Mohana was hearing a public interest litigation filed by advocate Narendra Kumar Goswami.

Goswami sought directions requiring the Centre to disclose all existing and proposed high-risk AI systems used by government ministries and agencies for welfare distribution, policing, surveillance and content regulation.

The bench observed that the petition raised highly technical questions that fell primarily within the policy domain and required the formulation of appropriate rules or regulations.

“We are not the experts. It is a highly technical issue and it is in policy domain,” the bench said, asking Goswami to submit his writ petition to the authorities as a formal representation.

The court noted that the petitioner had already submitted a representation to the government in February. It disposed of the petition without commenting on its merits and directed the authorities to examine his suggestions and consider appropriate measures.

Goswami was also given permission to supplement his earlier representation with further proposals.

The petition called for the establishment of a high-powered expert committee to develop binding guidelines on AI ethics. It sought mandatory algorithmic impact assessments and bias audits, greater transparency in AI-enabled surveillance and content moderation, human oversight of high-risk government systems and stronger data-protection safeguards.

It also sought public disclosure of a complete inventory of government AI systems used in areas including welfare allocation, predictive policing, facial recognition and online content moderation.

The petitioner further requested comprehensive impact assessments and independent bias audits of existing systems, with the results to be made public within six weeks.

Among its other prayers, the petition asked the Centre to pursue comprehensive parliamentary legislation regulating the development and deployment of AI by the state.

With PTI inputs