The Supreme Court on Thursday asked the Centre to consider suggestions made in a petition seeking binding guidelines on ethics, transparency and safeguards for the government's use of "high-risk" Artificial Intelligence (AI) systems in welfare, policing, surveillance and content regulation.
A three-judge Bench comprising Chief Justice of India Surya Kant and Justices Joymalya Bagchi and V Mohana was hearing a petition filed by advocate Narendra Kumar Goswami. The plea raised concerns over the absence of a comprehensive statutory framework governing the use of AI by government authorities, particularly in areas that can affect people's access to food, health benefits, wages, pensions, scholarships and welfare subsidies, PTI reports.
Court says AI issue falls in policy domain
"Tell us what can we do," the Bench told the petitioner at the outset.
Goswami said he was pressing one of the prayers in the petition seeking a direction to the Centre to file an affidavit disclosing all existing and proposed high-risk AI systems used by Union ministries or agencies in welfare, policing, surveillance and content regulation.
The disclosure sought details including the legal basis for the systems, their vendors and purpose, use of data, safeguards, grievance mechanisms and audit status.
"Send your writ petition as a representation," the Bench said, noting that the petition was comprehensive.
"We are not the experts. It is a highly technical issue and it is in policy domain," the Bench said.
The court observed that the issues raised in the petition required the formulation of an appropriate policy or regulation. The observation underlines the complexity of regulating AI in government systems, where technological questions intersect with decisions that can directly affect access to public services.
Petitioner seeks transparency and safeguards
The Bench noted that Goswami had submitted a representation to the government on the issue in February this year.
Disposing of the petition at this stage without expressing any opinion on its merits, the court asked the authorities to consider the suggestions and take appropriate measures. It also gave Goswami the liberty to supplement his representation.
In his public interest litigation (PIL), Goswami sought a direction to the Centre to constitute a high-powered expert committee to frame "binding guidelines on AI ethics, mandatory algorithmic impact assessments and bias audits, transparency in AI-based surveillance and content moderation, human-in-the-loop oversight for all high-risk governmental AI systems, and data protection safeguards."
The demands highlight a broader concern at the heart of the petition: when AI is used for decisions involving welfare, policing or surveillance, transparency and human oversight become especially significant because such systems can have consequences for individuals.
Also Watch:
Plea calls for audit of existing AI systems
The petition also sought directions to the Centre and others to conduct and make public, within six weeks, a comprehensive algorithmic impact assessment and independent bias audit of all existing AI systems deployed in welfare allocation, predictive policing, facial recognition and content moderation.
It further sought disclosure of a complete inventory of such systems, including their legal basis, vendors, purpose, data use, safeguards, grievance mechanisms and audit status.
Among its other prayers, the petition sought a direction to the Centre to take steps towards enacting comprehensive parliamentary legislation regulating the development and deployment of AI by the State, consistent with guidelines framed by the Supreme Court.
For now, the Supreme Court has left the policy response to the government rather than laying down judicial guidelines itself. The government's consideration of the representation will therefore be important in determining how concerns over accountability, transparency and safeguards surrounding high-risk governmental AI systems are addressed.
