Frontier AI Law: India's Two-Part Answer to Foreign Companies

 

Source:

https://www.ndtv.com/artificial-intelligence/india-weighs-us-style-frontier-ai-safeguards-as-government-prepares-new-law-12246199

India's coming frontier AI law must answer a question the NDTV report glosses over: 

how do you regulate OpenAI, Anthropic, and other US-based frontier

 models operating in India?

Direct legal control won't work. 

Foreign companies can't be forced to comply with Indian testing mandates or

 submit to certification. The NDTV report focuses on sandbox testing but leaves

 enforcement blank.

I've already outlined the solution in AIR Act - 2026: a two-part framework.

Part A 

regulates Indian companies directly—registration, certification before public release, strict liability.

Part B 

regulates foreign companies through market access conditions, not direct mandates. 

No company can offer a frontier model to Indian users without:

• An India Access Certificate (conditional on safety testing) 

• A resident Grievance Officer and Compliance Officer in India

• A security deposit to cover penalties 

• Clear acceptance of Indian jurisdiction

If they violate these conditions, India doesn't try to prosecute abroad. 

Instead, it uses graded enforcement: warnings → advisory → suspension →

 blocking under Section 69A of the IT Act, enforced at the ISP/infrastructure level.

Why This Matters

A frontier model causing harm in India can't be shut down by Indian police. But

 the platform offering it—whether App Store, payment gateway, or cloud provider

—can be. That's leverage.

The consultation paper must be explicit: 

offering AI services to India is a privilege, not a right, and it's conditional on

 compliance with Indian law.

---

With regards,

Hemen Parekh

www.Iam-Immortal.ai / www.IndiaAGI.ai / www.YourContentCreator.in