The Artificial Intelligence (Ethics and Accountability) Bill, 2025

Bharti Pardhi, a Bharatiya Janata Party MP from Madhya Pradesh, introduced the Artificial Intelligence (Ethics and Accountability) Bill, 2025 in the Lok Sabha during the 2025 Winter Session. MediaNama reported that although the bill is dated 20 January 2025, it was introduced on 5 December 2025, according to the Parliament website.¹ Its stated aim is to establish an ethics and accountability framework for the use of artificial intelligence technologies.¹

The bill's preamble, as summarised by MediaNama, cites the rapid expansion of AI as a driver of economic and social progress but warns that unchecked use raises ethical and regulatory concerns, including algorithmic bias, misuse of surveillance capabilities and a lack of transparency and accountability in automated decision-making.¹ It was introduced when the Union Government's public position was that India did not need a dedicated AI law. In December 2025 MeitY Secretary S. Krishnan said the government was not planning one, a position the minister later reversed (see Current status).²

The bill defines artificial intelligence as computer systems or applications capable of performing tasks that typically require human intelligence, including but not limited to decision-making, language processing and visual perception.¹ It then regulates the field through a single new body, a centrally funded Ethics Committee for Artificial Intelligence. According to MediaNama's summary, the Committee would have a chairperson, several representatives from academia, industry, civil society and government, and experts in law, data science and human rights appointed by the Committee.¹

The summary reviewed does not say which AI systems would be covered by a risk threshold, who counts as an AI developer, or how the bill would relate to the Information Technology Act, 2000 or the Digital Personal Data Protection Act, 2023. This entry did not review the bill's text.

Ethics Committee functions. The Committee would develop and recommend ethical guidelines for AI technologies, monitor compliance with ethical standards governing AI systems, review cases of misuse, bias or violations of the Act, promote awareness and build capacity among stakeholders, and carry out other prescribed functions.¹

Transparency and bias. Every AI developer would have to disclose an AI system's purpose, limitations, training data sources and methodologies, and the reasons for decisions that affect individuals. Developers would have to prevent algorithmic bias through regular audits and diverse and inclusive training data, withdraw materially biased systems until they are corrected, and keep records that show compliance with the ethical standards in the Act. The Committee would also have to ensure that AI companies use diverse and inclusive training datasets.¹

Surveillance and sensitive uses. AI systems used for surveillance would be limited to lawful purposes with the Committee's prior approval. The bill does not explain what counts as lawful purposes, which MediaNama says leaves interpretation to the proposed Committee. AI systems in law enforcement, financial credit and employment would have to be free of discrimination on the basis of race, religion and gender, and would be subject to stringent ethical reviews by the Committee.¹

The Committee would run a grievance redressal mechanism and have investigative powers so that users can file complaints about misuse or harm caused by AI technologies. The bill is also described as proposing an independent ethics committee and penalties for violations, with the stated objective that AI operates in the public interest and aligns with democratic values.¹

The Pacta repository describes penalties of up to Rs. 5 crore and criminal liability in severe cases. The MediaNama summary reviewed does not give penalty amounts, and the bill text could not be reached on sansad.in or PRS for this entry, so those figures are not verified here and are not relied on.

MediaNama's analysis said the bill does not offer a comprehensive treatment of AI issues. It said the bill does not provide a framework for fair compensation or royalties when data owners' data is used for AI training, does not say whether the Committee could inspect datasets when complaints arise about outputs or training data, and does not address accountability for copyright infringement. It noted that the DPDP Act allows AI training on personal data that is publicly available.¹

MediaNama also placed the bill against a backdrop of expanding state use of AI for policing, including Maharashtra's statewide rollout of MahaCrimeOS AI, and said that private member's bills have very low chances of debate or passage, although studying their policy questions remains useful.¹ The Union Government's position moved during 2026. In June 2026 Union IT Minister Ashwini Vaishnaw said a new AI law was needed, and on 8 October 2026 he said MeitY would release a regulation-focused consultation paper on AI within a month.²³

As of 15 October 2026 the bill is recorded as introduced in the Lok Sabha on 5 December 2025. The sources reviewed show no discussion, referral to a committee or vote, and this entry could not check the Lok Sabha or PRS records directly, so it cannot rule out later action.¹ Private member's bills do not become law unless the House passes them, and the sources reviewed give no sign of that. The Government's announced consultation paper on AI regulation, due about a month after 8 October 2026, is the separate track on which any AI law is more likely to move.³