Oversight and accountability
At the heart of the proposed legislation is the establishment of an Ethics Committee for Artificial Intelligence, a multidisciplinary oversight body comprising experts from academia, industry, civil society, and government. The committee would develop ethical guidelines, monitor compliance, and review instances of misuse or bias in AI systems.
The bill introduces explicit restrictions on AI-based surveillance, requiring deployments to be limited to lawful purposes and subject to prior approval by the Ethics Committee. AI systems used in critical areas such as law enforcement, financial credit, and employment would face heightened ethical scrutiny and would be prohibited from discriminating on the basis of race, religion, or gender.
Developer responsibilities
The legislation places significant obligations on AI developers. They would be required to ensure transparency by disclosing the intended purpose and limitations of their systems, the data sources and methodologies used for training, and the reasoning behind decisions affecting individuals.
Developers must also actively prevent algorithmic bias by conducting regular audits, ensuring diversity in training datasets, and withdrawing systems that exhibit significant bias until corrective measures are implemented. The requirement to maintain records of ethical compliance would strengthen traceability across the AI life cycle.
Enforcement measures
The bill establishes a grievance redressal mechanism enabling affected individuals or groups to file complaints with the Ethics Committee. Noncompliance may attract financial penalties of up to 50 million rupees, depending on the severity of the offence. Repeat violations could lead to criminal liability, reflecting a strong deterrent approach.
