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: