Artificial Intelligence has rapidly moved from being an efficiency tool to becoming a strategic business asset. However, as organisations increasingly rely on AI for decision-making, content generation, legal support and customer engagement, regulators across the world are introducing comprehensive governance frameworks to address concerns around transparency, accountability and data protection.
India’s proposed Regulations for Use of Artificial Intelligence in Courts, 2026, issued by the Supreme Court, are another significant step in this evolving landscape. Although directed at judicial institutions, the draft regulations provide valuable insights into the standards of AI governance that businesses, legal departments, technology providers and AI developers should begin preparing for.
Why This Matters for Business
Although the regulations are intended for courts, they reflect broader regulatory expectations regarding AI governance in India. Organisations that use AI-powered legal tools, document automation, contract review systems, compliance software or generative AI platforms should closely monitor these developments.
The draft reinforces several principles that are increasingly becoming global best practices:
- Human oversight over AI-generated outputs
- Transparency regarding AI usage
- Accountability for AI-generated errors
- Data governance and cybersecurity
- Regular auditing of AI systems
India’s ai draft rules
India’s draft “Regulations for Use of Artificial Intelligence in Courts, 2026” takes a very clear stance: AI can assist the justice system, but it cannot replace human judges. The proposed regulations are intended to govern the use of AI in judicial, adjudicatory and administrative functions of the Supreme Court, High Courts, tribunals and statutory commissions performing adjudicatory functions.
The draft regulations reinforce an important global principle—that AI may enhance professional decision-making but cannot replace human judgment in high-risk situations. Businesses deploying AI in sectors such as banking, healthcare, insurance, recruitment or legal services should similarly ensure that critical decisions remain subject to meaningful human review.
The draft makes it clear that lawyers or litigants cannot escape responsibility by blaming AI for false or misleading filings and the court may also pass such orders as it deems fit against the responsible person.
The framework places several absolute restrictions on the use of AI in courts. AI systems cannot decide case or judicial outcome, pass sentence, determine bail eligibility, assess flight risk or likelihood of reoffending, evaluate witness credibility, predict future conduct of witnesses or accused persons etc. That the ultimate authority to determine questions of law, fact and justice must remain exclusively with judges. Even where AI tools are used, accountability for decisions would continue to rest with the judicial officer or court official concerned.
The regulations propose establishing a new governance framework centered around a Permanent apex body at the Supreme Court level, which will facilitate safe and secure use of AI. This body would feature Supreme Court and High Court judges, technology and cybersecurity experts, finance professionals and tech-law experts, and would establish national standards, approve AI systems and issue governance reports annually. All AI systems would be subject to technical, legal and ethical audits at least once per year, and there would be an AI Register of approved tools and incidents in courts.
The draft regulations are also closely aligned with India’s Digital Personal Data Protection Act, 2023, and require compliance with applicable data protection and cybersecurity laws. Personal data cannot be used to train or refine AI systems without approval, sensitive judicial data gets heightened protections and cannot be transferred externally without authorization. No private entity can participate in AI systems for courts without prior approval; contracts must address data ownership, restrictions on using judicial data, disclosure and liability for AI-related harm, and courts must retain ownership or a perpetual royalty-free licence over tools developed using court data or resources.
For businesses and law firms increasingly using generative AI to prepare contracts, legal opinions, due diligence reports and litigation documents, the proposed disclosure requirements underline a broader expectation of transparency. Organisations can no longer assume that AI-generated work products are exempt from professional responsibility. Internal verification mechanisms will become an essential compliance requirement.
EU AI Act
EU AI Act: risk-based, extra-territorial and strict
While India is focusing on AI in courts, the EU AI Act takes a different angle with the same aim of safe, trustworthy AI. It applies to almost every part of the AI value chain – providers, deployers, importers, distributors and manufacturers – and even to companies outside the EU whose AI or its outputs are used within the EU.
The EU AI Act uses a risk-based model. It applies to almost every part of the AI value chain – providers, deployers, importers, distributors and manufacturers , even to companies outside the EU whose AI or its outputs are used within the EU.
The Act regulates AI by risk. High-risk AI systems include products or safety components covered by specific EU laws (such as toy safety and medical devices) and systems used in areas like employment, education, medical devices, elections, critical infrastructure management and biometric identification. In some cases, an exception may apply where the system does not pose a significant threat, however profiling systems that process personal data to evaluate behaviour or preferences remain high-risk.
Providers of high-risk systems must implement continuous risk management, strong data governance for training and testing data, measures to detect and mitigate bias Deployers must use systems as intended, maintain logs where possible and, for certain essential services, carry out fundamental rights impact assessments before first use.
The Act also includes transparency rules for AI that interacts directly with people or generates content. For example, chatbots must inform users that they are AI systems, and AI-generated text, images or video (including deepfakes) must be marked in machine-readable formats as AI-generated or manipulated.
There are also special rules for general purpose AI such as GPAI. Large model providers which can be used for many different tasks must respect EU copyright laws and publish detailed summaries of their training datasets. Non-compliance of the rules comes with serious penalties up to 35 million euros or 7% of global turnover for the most serious violations, higher than under the GDPR.The law entered into force on 1 August 2024, with bans on prohibited AI practices effective from 2 February 2025, GPAI rules from 2 August 2025, high-risk AI obligations from 2 August 2026 and full application, including certain product-linked systems and older GPAI models, by 2 August 2027.
US: A Patchwork of State and federal Rules
Unlike the EU’s comprehensive act, the US is moving through a mix of state laws and targeted federal measures. The Trump administration has called for Congress to pass federal AI legislation that would pre-empt burdensome state laws, but states continue to pass their own rules on issues like chatbots, pricing, deepfakes, high-risk AI and frontier models.
States like California and New York have started regulating “frontier” AI models – the most advanced general-purpose models. Large developers must prepare safety and security frameworks, report serious incidents and provide more transparency about risks and uses. In parallel, updated privacy rules in California demand notices, opt-outs and explanation rights when automated decision-making tools are used for significant decisions like credit, housing or employment.
Alongside these laws, US regulators repeatedly stress the need for internal AI governance such as clear approval processes for AI systems, incident response plans and careful external disclosures about where and how AI is used.
US states are also taking steps to deal with AI-driven pricing and content harms. Maryland has enacted its 2026 “surveillance pricing” ban for food retail and third-party delivery services, which prohibits dynamic/personalised pricing) that involves protected class information. New York has enacted the Algorithmic Pricing Disclosure Act, which mandates disclosures when personalised pricing is made based on personal information.
As of May 2026, the federal Take It Down Act will ban the knowingly publication or threatened publication of intimate images, including AI-generated deepfakes; covered platforms will also be required to remove content from their platforms within 48 hours and to address cases of duplicates. States like California have made it more explicit that AI-generated child sex abuse footage and other computer-generated explicit material is covered by criminal laws.
Business Takeaways
Businesses using AI should consider:
- Conducting an AI governance assessment.
- Identifying high-risk AI applications within the organisation.
- Establishing AI usage policies.
- Introducing human review mechanisms.
- Maintaining records of AI-generated outputs.
- Reviewing vendor agreements for AI liability.
- Assessing compliance with data protection laws.
- Training employees on responsible AI use.
Conclusion
The Supreme Court’s draft regulations represent more than a framework for judicial use of AI, they signal the direction in which AI governance in India is evolving. Together with the EU AI Act and the expanding patchwork of US regulations, they underscore a common expectation: AI must operate within a framework of transparency, accountability and human oversight.
For businesses, compliance is no longer limited to data privacy or cybersecurity. AI governance is emerging as a core component of enterprise risk management, corporate governance and regulatory compliance. Organisations that proactively establish AI governance frameworks today will not only reduce legal and reputational risks but also build greater trust among regulators, customers and stakeholders in an increasingly AI-driven economy.
Authors – Navya Saxena, Associate & Shagun Gupta, Intern.
