Principles, Not Statutes: How Legistify’s Akshat Singhal Recommends Preparing Legal Functions for AI Rules That Haven’t Been Written Yet

Akshat Singhal, Founder & CEO, Legistify

New Delhi [India], November 7: For enterprise legal teams, artificial intelligence has created an unusual situation. The technology is already entering contract review, legal research, document analysis and risk assessment, while the regulatory frameworks governing many of these applications are still evolving. Legal departments cannot afford to ignore AI until every compliance requirement becomes clear. Equally, they cannot treat the absence of AI-specific legislation as permission to deploy these systems without adequate safeguards.

The challenge is to build a legal function that can use AI responsibly today while remaining prepared for regulatory expectations that may emerge tomorrow. That requires a shift in thinking. Instead of organising AI governance exclusively around existing statutes, enterprises need to establish principles that remain relevant even as laws, technologies and business practices change.

For legal leaders, the question is no longer whether AI regulation will influence their operations. It is whether the systems, processes and accountability structures being established today will be capable of adapting when those requirements become more defined.

Build Governance Around Risks, Not Regulatory Predictions

One of the difficulties with emerging technology regulation is the temptation to predict exactly what lawmakers will require. Organisations may delay implementation while waiting for detailed rules, or build compliance processes around assumptions that could change as regulations develop.

Neither approach provides a dependable foundation for long-term governance.

A more practical starting point is to identify the risks that already exist when AI interacts with legal information. These include inaccurate outputs, unauthorised disclosure of confidential material, unclear accountability, insufficient documentation and decisions made without appropriate human oversight.

For instance, an AI system assisting with contract review may identify unusual clauses or suggest alternative language. The immediate governance question is not simply whether a future regulation will permit that capability. It is whether the organisation can establish what information the system relied upon, how its recommendations are evaluated and who remains responsible for approving the final contract.

Principles such as accountability, transparency, data protection, reliability and proportional human oversight provide a more durable foundation than attempting to anticipate every regulatory provision. They allow organisations to introduce controls that serve a practical purpose today while remaining adaptable to future requirements.

India’s AI Governance Guidelines: A Principles-Based Approach

India has already taken steps towards responsible AI adoption through the India AI Governance Guidelines, released by the Ministry of Electronics and Information Technology (MeitY) on 5 November 2025. Developed under the IndiaAI Mission, the guidelines establish seven guiding principles and six governance pillars covering areas such as accountability, transparency, safety and risk management. Rather than introducing a standalone AI statute, they advocate a principles-based approach that builds on existing legal frameworks while allowing flexibility for technological developments. For enterprise legal teams, these guidelines provide a useful foundation for establishing internal AI policies, strengthening oversight mechanisms and preparing for future regulatory requirements without waiting for comprehensive AI-specific legislation.

Treat Existing Legal Obligations as the Starting Point

The absence of a comprehensive, AI-specific statute does not mean that AI operates outside the law. In India, the Digital Personal Data Protection Act, 2023, the Information Technology Act, 2000, applicable contractual obligations and sector-specific regulatory requirements can already be relevant to how enterprises deploy AI systems.

The applicability of these requirements depends on the nature of the information processed, the organisation’s activities and the implementation status of relevant provisions. Legal teams therefore need to distinguish between obligations already in force, enacted provisions awaiting commencement and emerging policy proposals.

This distinction matters because AI governance cannot be built on regulatory speculation alone. Enterprises should begin by mapping where AI is being used, what categories of information are involved, which vendors or internal systems process that information and what existing legal duties apply.

A contract analysis tool handling confidential commercial agreements may raise different concerns from an internal knowledge assistant processing general policy documents. Similarly, a system supporting employment-related decisions may require different controls from one used to summarise litigation updates.

The objective should be to connect each AI use case with its actual legal and operational exposure rather than apply a single generic compliance checklist across the organisation.

Keep Human Accountability at the Centre of Legal AI

As AI systems become more capable, organisations may be tempted to measure progress primarily through the number of tasks automated. For legal functions, however, automation should not be confused with the transfer of professional responsibility.

An AI-generated clause can be grammatically convincing while introducing a commercial obligation that the organisation never intended to accept. A summary of litigation documents may omit a qualification that materially changes the interpretation of a dispute. A research output may appear authoritative even when the underlying sources are incomplete or outdated.

These limitations make human oversight particularly important in consequential legal workflows.

Legal departments should establish clear boundaries between tasks where AI can provide assistance and decisions that require qualified human review. Routine document classification, information extraction and preliminary summarisation may be suitable for greater automation, provided the systems are appropriately tested and monitored. Final legal advice, approval of material contractual positions and significant litigation decisions should remain subject to accountable human judgment.

The relevant principle is not that every AI-generated sentence must receive identical scrutiny. It is that the level of review should reflect the potential consequences of an error.

Make Traceability Part of Everyday Legal Operations

One of the most important capabilities legal departments can develop is the ability to reconstruct how AI-assisted work was performed. When a recommendation is challenged, an organisation should be able to identify the relevant source information, applicable workflow, review process and final decision-maker.

This requires more than maintaining technical logs. Legal teams need practical documentation standards that establish when AI was used, which outputs influenced a decision, what checks were performed and who authorised the resulting action.

Consider a contract negotiation involving several rounds of AI-assisted review. If a disputed provision later becomes the subject of litigation, the organisation may need to understand how that language entered the agreement and whether it passed through the appropriate approval process.

Building traceability into legal workflows from the beginning is considerably more effective than attempting to reconstruct decisions after a dispute or regulatory inquiry arises.

It also helps enterprises evaluate AI performance over time. Documented corrections, recurring errors and exceptions can reveal where additional safeguards, better data or changes to the workflow are necessary.

Prepare for Regulatory Change Without Rebuilding Everything

The regulatory landscape will continue to develop, but organisations should not need to redesign their entire legal technology infrastructure whenever a new requirement emerges.

This is where modular governance becomes valuable. Enterprises can establish common principles for data handling, vendor assessment, human oversight, access controls and recordkeeping, while allowing specific procedures to change as regulatory obligations become clearer.

AI vendor agreements, for example, should address confidentiality, permitted data use, security responsibilities, incident reporting and the allocation of contractual risk. Internal policies should define approved use cases, escalation procedures and review responsibilities. Periodic assessments should examine whether the systems remain appropriate as their capabilities and the organisation’s reliance on them expand.

Legal departments should also maintain a regulatory monitoring process that distinguishes binding obligations from consultation papers, voluntary frameworks and emerging international standards. Developments such as the EU AI Act and internationally recognised AI risk-management frameworks can offer useful reference points, but their applicability should be assessed rather than assumed.

The goal is not to build an unnecessarily complex governance structure. It is to create one that can absorb new requirements without disrupting the legal function’s ability to operate.

The Legal Function Must Help Shape Enterprise AI Adoption

AI governance cannot remain exclusively the responsibility of technology or information security teams. Legal departments have an important role in translating regulatory uncertainty into practical business controls, particularly when AI influences contractual obligations, commercial decisions, intellectual property, disputes and sensitive information.

That responsibility also requires closer collaboration with procurement, compliance, finance, human resources and technology teams. Legal leaders need visibility into where AI is being introduced, how vendors are evaluated and which decisions may carry material legal consequences.

For enterprises, this represents an opportunity to move beyond reactive compliance. A legal function equipped with reliable information, structured workflows and clear accountability can support responsible innovation without becoming an obstacle to adoption.

The most effective legal teams will not necessarily be those that predict the next AI regulation correctly. They will be those that establish governance processes capable of responding when regulations change.

AI rules will become more detailed, and some expectations will inevitably evolve. But the underlying responsibilities of an enterprise legal function will remain familiar: protect sensitive information, manage risk, preserve accountability and ensure that consequential decisions can withstand scrutiny.

The strongest preparation for tomorrow’s AI regulations is not a policy written around laws that do not yet exist. It is a legal function built around principles that remain defensible regardless of how those laws ultimately take shape.

By Akshat Singhal, Founder & CEO, Legistify