What Illinois AI Safety Law Means for India's Approach to Tech Regulation

## Two States, Two Approaches

Illinois passed the most comprehensive AI safety law in the US. India is simultaneously drafting its Digital India Act. The contrast in regulatory philosophy tells us everything about how AI governance will diverge globally.

## Illinois: Audit-Based Regulation

| Requirement | Mechanism |
|------------|-----------|
| Independent audit | Third-party access to AI systems |
| Whistleblower protection | Employees can report unsafe practices |
| Pre-deployment testing | External safety testing before launch |

This is a process-based approach: define what good looks like, require verification by independent parties, protect the people who report violations.

## India's Emerging Model

India's AI regulation is developing through three parallel tracks:

1. **Digital India Act**: The replacement for IT Act 2000, expected to include AI-specific provisions
2. **NITI Aayog's Responsible AI framework**: Voluntary guidelines (currently) that may become mandatory
3. **MeitY advisories**: Government advisories requiring "explicit permission" for untested AI models — a de facto licensing regime

## The Structural Difference

| Aspect | Illinois (US) | India (Emerging) |
|--------|-------------|-----------------|
| Approach | Audit + whistleblower | Licensing + advisories |
| Regulator | Independent auditors | Government (MeitY) |
| Burden | On companies (prove safety) | On government (grant permission) |
| Speed | Slow, thorough | Fast, potentially arbitrary |

## What India Should Borrow from Illinois

1. **Independent audit, not government licensing**: Government permission creates bottlenecks and corruption vectors. Independent audit creates a market for safety verification.
2. **Whistleblower protection**: India has no meaningful tech whistleblower protection. The Illinois law provides a template.
3. **Compute thresholds, not blanket rules**: Regulate based on model capability (compute thresholds), not all AI applications. A spam filter and a frontier model should not face the same regulatory burden.

## The Risk

India's instinct toward licensing (visible in MeitY's "explicit permission" advisories) could stifle India's AI startup ecosystem before it matures. Illinois's audit model — burdensome for startups but navigable — is more compatible with innovation.

What Illinois AI Safety Law Means for India's Approach to Tech Regulation

// COMMENTS

ON THIS PAGE