Use AI but retain final decision-making powers with humans: Parliamentary panel

New Delhi: Amid a "shift towards" data-driven and technology-enabled administration, a Parliamentary panel has insisted that an “ethical, secure and accountable use” of Artificial Intelligence (AI) should be ensured, with final decision-making authority be left to “designated human officers” and not machines.

The recommendation by the Parliamentary Standing Committee on Personnel, Public Grievances, Law and Justice came in its report on the Demands for Grants (2026-27) of the Department of Personnel and Training (DoPT) tabled in Parliament recently.

The panel headed by senior BJP MP Brij Lal acknowledged the deployment of AI-enabled dashboards, AI-based deputation alerts, AI-assisted recruitment rule generation and the introduction of AI-enabled tools at Lal Bahadur Shastri National Academy of Administration (LBSNAA) and other training institutions.

It said the initiatives taken by the DoPT shows an “important shift” towards data-driven and technology-enabled personnel administration. 

However, the panel said, given the “scale and sensitivity” of the government operations, there “remains significant scope” to institutionalise and mainstream the use of AI in day-to-day functioning, albeit in a “structured and accountable manner”.

It said the DoPT, in coordination with relevant ministries and departments, takes the lead in developing a comprehensive government-wide “framework for the ethical, secure and accountable use” of AI in public administration. 

“Such a framework should clearly define safeguards for data protection and confidentiality, ensure proper record-keeping of AI-assisted actions, and retain final decision-making authority with designated human officers. This is particularly important as AI tools across ministries may handle varied and sensitive categories of data, including personnel, financial, regulatory and citizen-related information," it said.

It also recommended that the use of AI tools for official purposes be governed through centrally approved enterprise-level agreements with service providers, including Large Language Model (LLM) platforms, so as to ensure confidentiality of government data, compliance with national data protection and security standards, and clear contractual safeguards regarding data storage, processing and access.

The AI-enabled systems adopted by the ministries adhere to robust information security protocols and confidentiality safeguards, particularly in handling service records, vigilance matters, performance data, financial transactions and other sensitive government information, it said.

It also emphasized the need to expand AI literacy and responsible use training across all levels of government, particularly for dealing hands and lower-level staff who routinely process official data, so that technological adoption is accompanied by awareness of ethical boundaries, confidentiality obligations and cybersecurity safeguards.