Protecting Global Citizens by Securing Communications with AI

Abstract

The panel explored how AI‑driven security can shift the burden of fraud protection from individuals to a “smart” network core. Participants debated the regulatory, technical and business challenges of deploying AI at national scale, covering risk perception, the need for a nimble regulatory toolbox, the role of the judiciary, data sovereignty, sandbox‑based innovation, and the trust gap between legacy payment infrastructure (UPI) and emergent AI systems. An audience Q&A highlighted India’s sovereign data‑asset strategy, incentives for data‑center investment, and practical pathways for cross‑regulatory sandbox experimentation.

Detailed Summary

  • Moderator (Sanjeev Sir) opened by questioning whether the panel was under‑estimating risk in the rapid rollout of AI‑enabled security solutions.
  • Vikram Sinha responded that zero risk is unattainable; the focus should be on equipping organisations with tools that can adapt as technology evolves. He highlighted a regulator’s need for nimble, up‑datable frameworks (“the tools are there, we just need to use them effectively”).

Key Insight: Risk cannot be eliminated, but agility in governance can mitigate emerging threats.

2. The Judicial System as a Back‑stop for AI‑Induced Disputes

  • The moderator turned to “Sanyal Sir” (likely a senior regulator or legal expert on the panel) to discuss a structural reform India must prioritize.
  • Sanyal Sir argued that the judicial system must be accelerated because AI will create novel disputes that existing law cannot anticipate (e.g., liability for AI‑driven decisions, ownership of AI‑generated inventions).
  • He posed several hypothetical scenarios:
    • An AI model produces a harmful outcome – who bears the damages?
    • In copyright contexts, does the prompt author, the data‑owner, or the algorithm own the resulting innovation?

Recommendation: Build a judicial pipeline capable of fast‑tracking AI‑related cases, recognising that new “philosophical” problems will surface quickly.

3. Trust, Transparency and the Analogy with UPI

  • A panelist (likely Anshuman Kar or Uday Reddy) likened AI to the invisible layer behind India’s Unified Payments Interface (UPI), which was initially praised for its trust in a black‑box system.
  • Sanjeev Sir contrasted UPI’s non‑emergent, deterministic nature (“send 100 ₹ → receive 100 ₹”) with AI’s emergent behaviour—outputs can vary over time, sometimes degrading or even lying.

Takeaway: While AI can drive innovation, blind trust is unsafe; skeptical, bounded‑problem approaches are needed.

4. Audience Question – Sovereign Data Assets & the Indian Stack

  • Aditya, founder of First Tile, asked for the panel’s view on India’s sovereign data‑asset strategy.

    • He highlighted the budget‑announced tax holiday for data‑center construction, framing data‑centres as the “oil rigs” of the new data‑oil economy.
    • He warned that data ownership rights must accompany data‑availability; otherwise, India would have resource without the means to refine it.
  • Panel response (Neha Gupta Mahatme & Bipin Preet Singh):

    • Emphasised that large‑scale LLMs are only one slice of AI potential; many bounded‑problem use‑cases remain for SMEs and startups.
    • Highlighted the government‑backed ecosystem (budget incentives, sandbox programmes) that lowers the barrier for innovation without massive capital.

Key Insight: Data sovereignty coupled with regulatory incentives can trigger a vibrant, home‑grown AI ecosystem.

5. Sandbox & Inter‑Regulatory Collaboration

  • Praveen (likely a regulator) outlined the interoperable sandbox that links IFSA, RBI, SEBI, and IRDAI.
  • Murali clarified that while the sandbox is technically ready for any product, it is only invoked when a regulated entity believes a rule is being breached.
  • The panel discussed the legal mismatch between domestic and offshore contexts (e.g., INR vs. foreign‑currency transactions in IFSA), illustrating why legal harmonisation is the primary barrier, not technology.

Recommendation: Expand sandbox scope to support innovation (compute, data, tools) while ensuring legal compatibility across jurisdictions.

6. Practical Steps from the Ecosystem Perspective

  • Bipin Preet Singh (Mobikwik) shared the First High experience: building a customer‑data platform that aggregates synthetic data and plans to correlate consent‑based data from regulated entities.

  • He suggested a “consent‑backed API” for data consumption, urging regulators to define standards that give data processors a seat at the regulatory table.

  • Anshuman Kar and Uday Reddy highlighted their participation in previous sandbox hackathons, praising the RBI‑in‑tech engagement and the rapid prototyping it enables.

Open Question: How can regulators institutionalise such collaborative sandboxes beyond ad‑hoc events?

7. Closing Remarks – Embedding Governance in AI

  • Sanjeev Sir reiterated the core theme of the summit: People, Planet, Progress.

    • People – protect citizens from opaque AI decisions.
    • Planet – scale responsibly and sustainably.
    • Progress – pursue fair innovation, not just speed.
  • He warned against over‑reliance on emergent AI (“do not trust it”) and advocated for bounded‑problem deployment with continuous verification.

  • The moderator thanked the panel and announced that Krithi would present moments from the India AI Mission, signalling a governmental commitment to the discussed agenda.

Key Takeaways

  • Zero risk is impossible; organisations must focus on agile governance tools that evolve with AI capabilities.
  • India’s judiciary requires acceleration and specialization to address AI‑generated disputes (liability, IP ownership, etc.).
  • Trust in AI must be conditional; unlike deterministic systems such as UPI, AI can produce variable, occasionally erroneous outputs.
  • Sovereign data assets are a strategic national resource; ownership rights and processing capabilities must accompany data‑center incentives.
  • Bounded‑problem AI applications (e.g., fraud detection, credit underwriting) are the low‑risk entry points for startups, not just large LLM projects.
  • The interoperable sandbox across IFSA, RBI, SEBI, and IRDAI offers a sandbox for innovation, but legal harmonisation remains the main bottleneck.
  • Consent‑backed APIs and clear regulatory definitions for data processors can bridge the gap between regulators and innovators.
  • Embedded governance—continuous monitoring, verification, and a culture of healthy skepticism—is essential for fair, sustainable AI progress.

Prepared by the AI Conference Summarisation Team.

See Also: